Cadmore media player playing video From Manuscripts to Machines: A Fun Dive into AI for Publishing Newbies
Video Player
Accessibility
Using forms mode, you can utilize keyboard shortcuts to make usability more efficient.
Space bar will play and pause the video. Pressing escape will close all displays and pressing it again will jump to the end of the player.
You can press A to open the Keyboard Shortcut Dialog which also provides an audio description of the player and all of its features.
Language
1x
Play Speed
Adjust Volume
Transcript
Search
Result Count
of
Click to jump to result
Search
Re-center
Segments
Result Count
of
Click to jump to result
Search
Name:
From Manuscripts to Machines: A Fun Dive into AI for Publishing Newbies
Description:
From Manuscripts to Machines: A Fun Dive into AI for Publishing Newbies
Thumbnail URL:
https://cadmoremediastorage.blob.core.windows.net/2e42c526-f19f-4626-a024-0baf70e24bfb/thumbnails/2e42c526-f19f-4626-a024-0baf70e24bfb.png
Duration:
T01H01M01S
Embed URL:
https://stream.cadmore.media/player/2e42c526-f19f-4626-a024-0baf70e24bfb
Content URL:
https://cadmoreoriginalmedia.blob.core.windows.net/2e42c526-f19f-4626-a024-0baf70e24bfb/GMT20250219-155906_Recording_gallery_1920x1080.mp4?sv=2019-02-02&sr=c&sig=zin0zGspKtQ8WexAhjt4lLAVNpdbJGhUAGElCWNuu8o%3D&st=2025-02-23T06%3A27%3A39Z&se=2025-02-23T08%3A32%3A39Z&sp=r
Upload Date:
2025-02-23T06:32:39.4575868Z
Transcript:
Language: EN.
Segment:0 .
Thank you, and welcome to today's SSP early career subcommittee webinar, from manuscripts to machines a fun dive into AI for publishing newbies.
We are very pleased that you could join us today. My name is Rebecca Benner and I'm a member of the SSP early career subcommittee. I have just a few housekeeping items to review before we get started. Attendee microphones have been muted automatically, so please use the Q&A box to enter questions for the panelists. We would love to hear your questions and have set aside time at the end to address them.
You can also use chat to say Hello, or to alert us to any technical problems you are having, and we encourage you to introduce yourself in the chat and tell us where you're from. Closed captions have been enabled. You can view captions by selecting the More option on your screen and choosing show captions. This one hour session will be recorded and made available to all in the next few days.
A quick note on SPS code of conduct and today's meeting. We are committed to diversity, equity and providing an inclusive meeting environment, fostering open dialogue free of harassment, discrimination and hostile conduct. We ask all participants, whether speaking or in chat, to consider and debate relevant viewpoints in an orderly, respectful and fair manner. It is now my pleasure to introduce our speakers today.
Josh Nicholson is the chief strategy officer of research solutions and co-founder of scite, a next generation citation index of citations into context. Previously, he was the founder and CEO of the winner, winner and CEO of authorea, two companies aimed at improving how scientists publish and collaborate. He holds a PhD in Cell biology from Virginia Tech, where his research focused on the effects of aneuploidy on chromosome segregation and cancer.
Rocio Vidal is publication data and ethics manager at the American Chemical Society. Rocio joined ACEs in June 2024 as a publication data and ethics manager. In her role, she provides guidance to ACEs journals on ethics cases, promotes best practices, and helps DraftKings policies in accordance with industry standards set by the Committee on publication ethics, for instance, and assesses research integrity tools for the ACEs publishing integrity office.
Prior to joining ACEs, Rocio worked as an editorial manager at science advances at ars. Hongzhou is the senior director of AI product management at Wiley, where he designs AI strategy, drives the product roadmap and leads teams in developing AI powered solutions that automate and enhance research and publishing processes.
Huang is a recognized a thought leader in the scholarly publishing industry. He articulates the vision, challenges, and opportunities of AI and transforming the field. Huang has been a Scholarly Kitchen chef since 2023, co-chairs alpsp, a special interest group, and also serves as a cope advisor and is a member of STEM future lab. Thank you Josh, Rocio and Hong for joining the panel today.
Before I turn to our panel discussion, we have a poll we would love for you all to complete. It should come up on your screen shortly and we would love to know. Do you use AI in your professional lives and work? And do you think you are using AI more or less than your colleagues?
It looks like most have answered. All right. Great Well, Thanks for Thanks for sharing your experience with AI. It's great to have a sense of how you all feel and what your experience looks like. So as we get started, feel free to share in the chat your name and location.
If you have questions for the panelists. Please enter them in the Q&A. All right, so my first question for the our panelists is, could you describe your pathway into publishing and how it led to your involvement with AI tools or ethical issues related to ai? I'll start with you, Josh. Yeah, Thanks, Rebecca. Hi, everyone.
It's nice to see the location. I thought that was fun. And so as you introduce me, my background is not in AI. My background is in Cell biology. As a PhD student, and even before then, I've always been interested in how science works. And so how do we publish, you know, how do we fund science, etc.? And so I used to just write these kind of pieces as a student that I would publish, analyzing citations, analyzing how we fund things slowly wanted to go beyond just publishing and to try and build some of these solutions.
And so the winnower was my first Foray into basically trying to fix everything I saw wrong with publishing. And so very naive and idealistic, you know, PhD student, not completely crazy though, so we started to publish things first and then winnow the good from the bad via post-publication peer review. So we started at the same time as like biorxiv ahead of my arxiv.
We were mentioned in all those different things. And it's funny, because if I look back on my emails from like 2013, I have emails saying, hey, we should start a archive. I spelled it differently than the current one. And so that's kind of how I got into, I would say, you know, publishing, if you will, was trying to launch an alternative, you know, publishing Foray from that.
You know, the winnower was not just aimed at trying to like be a new publisher, but was really trying to address the challenges of reproducibility. And so that is where site fits in. And, you know, it's funny because on the winnower, we published a paper suggesting this new way of looking at citations. Right wouldn't it be great if you could see if a citation was not just, you know, 1 to 1, but exactly the type of citation, how and why and where that citation came from.
And so started very manually analyzing hundreds of papers and looking at sentences where the citations were made realized you could break down and bucket citations into different types. And of course, there's literature looking at citation ontologies, but realized also, if you want to do this at scale, you need AI, right? You need some solution, right? Because we're talking about billions of sentences from research articles.
And so we I've been trying to talk to anyone that would listen to me to say, hey, you got this idea. Can we automate this for a long time? And I think we started building website at the right time because there was enough advances in AI to do what we do, which is classifying a citation statement as supporting or contrasting. And to do that, you know, with some level of accuracy that is acceptable.
And so that's where we are. You know, we spent a lot of time partnering with publishers, both publishers actually on the call to help build out these next generation citations and display them on the version of record. And I would say that is, you know, and this will maybe get to something we speak later. Kind of the more vanilla view of AI, right? We do use machine learning and deep learning to do this.
But, you know, historically that was not anything to do with generative AI. So that's kind of how I got into it, where we are with AI, and also why we started to use AI, not just because it was there, but to solve a problem. That's great. Thank you so much. Rocio, how about you?
All right. So I have a pretty different, I guess, upbringing through my career. But basically I did also start in a graduate degree. I very quickly into the graduate program decided that I didn't want to be a researcher myself, but I did want to contribute in some capacity to the scientific community. So I found myself applying to anything science related.
I ended up getting an entry level editorial coordinator job at a journal where I learned everything from, you know, submission through publication, like handling the peer review. Working in editorial journal. Operations, et cetera. And I agree with you that role into management of editorial staffs. And then, you know, early to 2021, 2020, when we were starting to see the uprise of the distrust in science through, you know, the pandemic, but also the growth of paper Mills, I started to get more and more interested in these ethical questions and dilemmas that we were facing as an industry.
And what better way than to learn more than just become like, have that become part of my role? So I decided to take on this new role as a data and ethics manager. And I've been, Yeah, learning a lot and growing a lot in the ethics space, especially when it comes to AI, but also other issues that we're facing as an industry. Yeah that's great.
Wonderful and Hong, how about you? Right I'm also different from different path. So after my working as a software engineer, lead and MBA. So I joined the Wiley and as a senior product manager for the information discovered in the 2017. So right when the deepmind the alphago made the history. So at that time. So with my PhD, because my PhD is in the 3D modeling with AI and the MBAs in the digital transformation.
So I knew I will be the future or the future. So I started building the why join the company and start building the AI teams, including the R&D and engineering teams. And also at that time, because the most of the customer people don't have the, you know, the confidence for the AI and it's still early. So we have to, you know, the identify the every possible the opportunities, build the prototype solutions to prove the value.
It's not easy. But you know, we're making progress. So and also that I'm very happy and feel the I'm glad in the I work in the scholarly publishing industry because this is a unique place industry for industrial Ii because our content, you know, the few, the millions, millions of content fields I training and also the I intern the enhance everything from authoring peer review to publish and discovery.
So this is a unique place to work on the AI. Yeah Yeah. So that's great. I think that leads really well into our next question, because I think it might be helpful for our attendees to sort of be clear about what we're talking about when we're talking about AI. And so maybe I'll start with you. Could you provide some context related?
You know, you were just mentioning, you know, our day to day activities and generative AI. So could you sort of level set for the our attendees? Yeah, I think for me it's a 2023. You know, two years ago 2023 is a clear the milestone before 2023. AI in publishing was mostly a task specific driven rebuilding the machine learning model. Deep learning model.
As Josh mentioned for a specific task, we built this for the recommendation for classification, for speech recognition, et cetera but after the 2023 is end of the, because it is the end of December of 2022. So Thanks to GPT and the large language model. So we have shifted from to a more unified approach. I mean, now a single large language model can handle translation, summarization or text generation and even complex Q&A tasks.
Question answered, you know, task which which is a very hard problem, which is almost impossible for the, you know, the individual, you know, the machine learning or deep learning model to do. But now it's possible. So the entry barrier technology barrier has become lower. So it makes you know, the product and the platform which integrates this AI technologies even more important already. That's great.
Josh, would you like to add to that? Yeah I mean, I would really just echo that. I think you captured it very well, right? We have been using AI, you know, for some time. And you can find Eugene Garfield talking about AI in the 60s, right? And so AI has been around. I don't think it's been appreciated because it was very task specific.
It wasn't so easily accessible. And I really think this 2023 was a defining year where chatgpt came out. You know, I think it's the fastest growing product ever, and we can all appreciate it. And I think that's why we're all talking about AI nowadays, right? Because AI has been around. We've been talking about it.
Theoretically, people are doing PhDs on it. But here is this kind of transformative moment that, you know, is just this powerful tool, very easily accessible, that you can plug into different systems and tools you know, quite easily and experience it. Right before you had to kind of be an engineer, understand this, and now you can go to chatgpt or any of these things. And you feel kind of that difference there. And so it's really kind of this, this big inflection point.
And I think it's interesting. But I do think it's important to kind of consider both. Right because it's so new we can be so scared of it. But there's also opportunities. And we have been using variations of AI for quite some time. Great Thank you. Rocio, how about you? Would you like to add anything?
No I mean, I would just continue to echo what it's already been said that really generative AI has brought up a disruption to the industry and all industries, really, and that we now have to consider things that we haven't even considered from a philosophical or an ethical point of view. Like what? How do we use it? Does it replace a review or does it replace an editor? How does our role change now that we have access to these tools, and how do we do that responsibly?
So Yeah, it's a very interesting time. Yeah, absolutely. That's great. So this is for everybody. But could you share your perspective on what types of AI are having the biggest impact that you think on scholarly publishing and what those impacts are? And maybe I'll start with you, Rocio, this time. Sure so, I mean, I already touched on my previous answer, but thinking about redefining what our role is as we use this tool, where are the guardrails for using it?
How do we define the use? Why do we, as an industry decide it's acceptable versus unacceptable use? And what impact is that going to have on how science evolves moving forward? I mean, when you think about AI and I'm sorry, science and reproducibility it like AI right now, at least to me, I'm not an expert on it, but it's very much a Black box where you feed something into a large language model and it spits something else out.
What's actually happening in the background? How do we build scientific content that's reproducible? So that is, I guess, questions that are going to have to be answered. And you know, as an industry, we're moving in the right direction. But still there's a lot to discuss and to define. Yeah that's great. Thank you.
Josh, how about you? Yeah, I think Rocio has got a tough job because ethical challenges, legal challenges to me are kind of like the biggest impact, if you will. Right? the technology is so far ahead of everything that people are using it. We don't know what to do with a policy whether, you know, we're at a journal or a University, and then again two weeks later, like there's some bigger advance, right?
And so the technology is moving at such a rapid rate that it's very hard to do anything from a legal or ethical or like pedagogical perspective, because it just changes so dramatically. And I do think that's where it's going to be maybe the most critical. Right can we use this for authoring? Right people are using it already. Is it enough to put a checkbox and say don't use it?
Well, as we see, like it's pervasive, they're going to use it for peer reviewing. We're telling them not to. But again, we need to kind of figure out the ethical, you know, applications and policies that we as publishers, we as teachers, we as researchers have because it's here. Right and people will use it. It does save time.
I mean, it's pretty amazing for certain things. But we need to recognize the risks, the challenges, the legal things. And so, you know, also what Hong said, I think it's a privilege to kind of be in our space of, of research, publishing, because LMS have ate up the entire part of the web, if you will. They haven't really touched research articles in any way.
And I think research articles, if you go back to the fundamentals, like are very valuable. And I don't mean that from just a monetary perspective, but from the content they have, right? They touch every aspect of our life. There's research on how Peppa Pig influences your children learning English as a second language to prostate cancer, to particle physics. They are backed by evidence, statistics, peer review.
And so how do we like how do we leverage these things together. Right this trust. And I think scholarly publications and even citations which I'm biased, obviously by that have a role here. And it's a very important role because LMS are hard to trust. And so, Yeah, it's interesting. But to me, I think the biggest impacts are going to be like some of these legal challenges.
And then also, how do we responsibly enter this, this post 2023 chatgpt world. Yeah, absolutely. I think, you know, those are some of the conversations we're having at our organization. You know, how it you know, it feels like there's some tasks that really that could utilize AI and really save some people time. But then there's we definitely keeping the legal and other aspects at the front of mind too.
So it's sort of being careful about that is definitely something we've been doing for sure. How about you, hon? What would you like to add? Right I think I try to, you know, the talk of something, some of the type of I from the mainly from the technology and the solution application perspective. So for me, there's several, the type of eyes I believe that have the biggest impact has already happened.
For example, the multi-modal the large language model, the AI which the like the GPT four. So it's not only the handle the generate the tax understand generate tax but also the handle the images, audio, video et cetera seamlessly. What does it mean to the research? Research means we can help the researcher to generate the much richer, you know, the input, research input and output.
So in the future it's not only the paper, not only maybe text based paper, but maybe, you know, the audio, video, and the corresponding images and the kind of stuff and Ar VR in the future, of course. And another is the right pipeline retrieval, augmentation. The generative pipeline, which is almost is a foundation for almost all the, you know, the discovery system today.
I mean, the generative AI based, the discovery system today is a conversational discovery. So to make sure that people not search for the keyword by keywords, search for keywords, return 100 results. You still need to find answer by themselves, but you can like talk to the experts and then the iteratively and until you find the answer and they get answer directly from the, you know, the millions of, the documents. So I think this is and also in the future I, I assume you know, the most of the content on the internet will be the AI generated in the future.
And another is the AI agent. So this will be very popular where everyone talk about. So this AI system which is at autonomously the break down the tasks, search for data, for example, the search for data. Use the tools intelligent so user can just specify their goals with agents. They can automatically break down the goal into the different tasks and decide what and where to search to, what to do, which tool to use.
So this is very, very useful. And and another thing is not about AI application tools itself, but about AI governance platform. So because as AI grows and the risk also grows of course. So they have these tools we need. We need to know the specific dedicate the tools, many governance tools to help us organize and manage the risk, comply with the regulations because the different, you know, the maybe different the country, the area have the different regulation for AI.
We need to make sure the align the our, the AI development, deployment with the best AI practice with ethical standards, et cetera by embedding the governance mechanism through the whole the AI life cycle, not just the AWARE when you use it, but to start when you have this idea, when you start building this and when deploy it or you, you just try to, you know, buy from the third party vendor or something. So all these needs of aligning with, you know, the ethical standard.
Yeah that's great. Good considerations. So let's move on and talk a little bit about opportunities and threats. I mean I think it's sort of a thread through all, you know, everything we've been talking about so far. But you know, as a specific question. Could you talk about some of the opportunities and, and threats that are attendees should be or aware of now or maybe be thinking about for the future?
So I'll start with you, Josh. Yeah I. Where should I start? So I think in terms of opportunities, I think there's a lot of uncertainty. Right and in uncertainty there's opportunities as well. And so to me I do think AI is not going away. And so we need to kind of come to the table as like what does that mean if it's not going away. How is this going to be used in peer review.
Not is it going to be used in peer review? And you know, from the perspective of starting companies, I do think there's interesting things to test where it's like, maybe this is not going to be the standard right now, but we can set aside, run some experiments, test different models, whether that's peer review AI systems, whether whatever it is to test, get numbers, look at this and understand here's a world with AI natively.
Right and I think there's opportunities for people to do that, you know, by starting companies to test individually. Because I do think New workflows are going to emerge out of this. Not all the workflows that we might love, which is maybe a threat, but there's going to be changes and we have to kind of come to the table with a sober view of, here is this powerful tool. It's going to keep getting more powerful.
What does that mean? Right that improve some of my day to day workflows? You know, it improves my emails. I draft an email and I'll say fix the grammar. Like make this more convincing. And it's what I had written. But then it like removes all the grammatical errors and the weird, you know, things that I had. And so I think it's, you know, we all have different kind of perspectives.
We're all doing slightly different things, probably in our day to day. And thinking about it from an n of one perspective can help, but then also thinking about it from like a perspective of like the industry. Is there opportunities that I see that make sense, that I think will become mainstream in the future? And going back to like my first introduction, like preprints weren't a thing in biorxiv, right?
You know, or in biology. Right they became a thing. Right? and I think starting to think of that from the AI perspective is what is going to be kind of the archive of I, if you will. Right and is that an opportunity for a publisher? For an editor? And I, you know, as a scientist, there's this saying, and I won't say it here because there's some curse words in it, but it's like, do the experiment.
Right test it. Right go test these things because everyone's going to have different opinions and thoughts. But ultimately, and this is why research is so amazing and why, like I am in this space, we need to test these things. How does this impact pedagogy? How does this impact peer review. How does this impact writing. And from that, we can make better policy, ethical decisions, et cetera.
But until then it's a lot of different shouting and me waving my hands even more than I'm waving them now. Yeah, I think that's a really helpful way to think about it. Approach it. You know, that's it's almost in a way, you know, like how we approach anything else. It just feels very large and impossible, maybe. But, you know, if we approach it using the same methods we've maybe used for other things, we can break it down and figure it out.
Maybe Yeah. How would you like to add? As you think about opportunities and threats that our attendees might want to know about. Sure to build on what Josh just mentioned, I think the AI is definitely is a very useful tool to increase the, you know, the productivity in our daily work, you know, work life or life. I mean, the for example, I think maybe the many people are already used, you know, the Google, you know, the Gmail, you can they provide based on the Google Gemini, the model, they can automatically draft help you to draft the email or something.
And also in the teams meeting Microsoft Teams meeting before we can just we can either accept or decline the meeting event. But now we can follow. We can follow the meeting. So there's a third options. So even if you don't attend the meeting they will still send you the record and the transcript automatically generated by AI and even the summarization.
You know the summary and also action point, which is very useful. So this is already the changed the way of our work. And from the research perspective, for researcher itself, I think the I can be used will be in the future. You know, the imaging of ais and not only the fund organized information, but also the think and the Apply the knowledge and help them plan a executed, you know, the research to work faster.
So it's basically it's like the personal the research assistant. This is, today in the Google. Apple already have the general assistant. But in few future we should have the is happening now. So we should every researcher should have a research assistant to help you to speed up the research work. So this is another thing another two opportunities. One is the AI for peer review.
Because one of the top challenges in the scholarly publishing is a slow, as you know, the slowdown, slow of the, you know, the peer review process. And then how can we apply AI to speed up this? So I can help you find the relevant reviewers from the, you know, the many, the 100, thousands of the, you know, the revered author database. And can we even generate the initial report review report based on the readability, novelty, reproducibility?
This is also happen. We can see, there are many more and more. The you know, the startup and the company is coming into to touch this area. And another area is AI for integrity. This is a bit special. This because it has both opportunity and the threat. So opportunities we can apply AI to better detect this the fraudulent papers, the paper mill papers is the AI generated content is in the relevant scope, in the scope of journal, et cetera but also the well the AI boost, the search quality.
It's also the integrity concern itself. So plagarism image manipulation paper mill. So which all this makes it much easier to generate all this. You know, the fake papers, fake images by using AI. So this so that makes detecting the AI generated fraudulent will be even harder. So this is an ongoing battle. Yeah, absolutely. And that's a great segue to Rocio.
You could talk about your perspective working in ethics. And what you see as opportunities and threats. Well, I'll start with a positive side on the opportunities. I think AI often opens a lot of doors for access, both for the authors that are, you know, maybe using it to improve their language, to improve their writing, to make it more convincing, like there is already data to show that a quality, high quality written paper improves your chances of getting accepted and published.
So I think it's going to open a lot of doors for non-native speakers and people that just struggled with the writing aspects of scientific work to contribute to the conversation in terms of, you know. Lessening the workload of reviewers. We also are already seeing tools that are helping identify new reviewers that an editor might have not even considered, or potentially like find ways where an I can focus on whether or not the paper is reproducible and that the reviewer, the peer expert, talk about potential ways to improve this research or to like, you know, find new directions to like, take the paper forward.
But then also on kind of like the, the access to the like content that we're publishing, AI offers a great opportunity to create plain language models that, like patients, for example, can use as they're like considering treatment methods or like generally people just are interested in the scientific content that are currently, you know, there is a language barrier, scientists write, for scientists, and the general public can't really access that very readily.
So I just opens that door to a broader audience. On the risk side, I mean, I wouldn't work in scientific publishing or, sorry, industry. The integrity side of it without talking about fraud. And like Hong said, it is AI is giving a very helpful tool to fraudulent content and people that want to create fraudulent content. So when we used to think about fraud five, 10 years ago, we were thinking about plagiarism, about someone opening Photoshop and drawing on their images or like beautifying them.
Now, why would you go and plagiarize someone when it's easier to just ask an AI bot to create an entirely new paper for you that gives you new text, new images, new video content? And like Han said, it is very difficult to police this and catch these as we and our vendors are like developing tools to catch and detect fraud. The paper Mills are also developing tools to circumvent those tools that we have access to.
So it is an ever evolving landscape. And, you know, we're there's very smart people working to create tools and to, you know, make advancements in that space. But it's also a growing and continuing challenge there. So we'll see what comes out of it. Yeah, absolutely. Before we move to our next question, there is a question from one of our attendees.
And I think it probably fits well with what we're talking about. But would the publishing industry move towards a standard i.e. usage practice, and if so, what can we expect? Interesting question. And I would like to answer what can you elaborate the more what's the standard i.e. usage practice?
What do you refer to? So maybe maybe he can clarify or they can clarify in the Q&A or in the chat. But we can also come back to it. So maybe if you could clarify that in the chat, we'll come back to your question. But for the next one then could you all talk about how you use AI in your daily activities.
So more from a practical standpoint, you know, a use case or, you know, it can be something as broad as that or just really just nitty gritty. How do you use it daily. And I'll start with you Josh. Yeah so I mentioned one kind of thing that I do, which is just cleaning up emails and trying to do some like data analysis with it.
And so I'll instead of like using Excel I guess, or like going to grammarly, I'll just copy and paste and say, here's my draft that I wrote, fix it. Like make it more convincing, make it nicer, whatever I want. And so I do use it in that way, and I think it's pretty powerful, right? It like, makes my email a lot more easy to read right to Rocio's point, it's helping, right? I am an English like native English speaker, but I don't have the best grammar.
And so that, you know, improves that. And I think I do use it kind of day to day looking at different data analysis, just like kind of back of the napkin math that I used to do. Now I'll have chatgpt kind of do for me, you know, beyond that, like we are seitai before LMS exist, right? So we are always using AI for a variety of different things. And I do think, you know, we're mostly talking about LMS here, but, you know, we use AI to classify citations.
This gets to integrity, reproducibility and looking at all these things I think it also helps with discoverability as well. And so we're constantly working with publishers to think about, you know, how can this help from kind of the product perspective, but then also the licensing perspective. And so scite joined research solutions, which was formerly preprints desk about 14 months ago.
Research solutions is doctor'll write delivering documents, you know, supplying the access and rights to articles on an a La Carte basis to large corporations. We think a lot about how can we partner with publishers to get some clarification on AI rights. Right is there a mechanism that allow you, allows you, as a publisher, to bridge this gap from like OpenAI to publisher, where there can be some use, right. There can be some control, we can answer some of these tough ethical, legal, monetary perspective questions.
I wouldn't say we've solved all of that, but I think, you know, we are partners to publishers. We're independent from publishers. And we think about it from the product perspective, the editorial perspective, and then the licensing and rights perspective as well. And so constantly thinking about it and then, Yeah, you know, just using it kind of maybe in the way that everyone kind of uses it to help with little tasks here and there.
Yeah Yeah. That's great. Thanks, Josh. How about you, hon? Right so I think that we can. I can try to put it into two categories. I for the daily work for myself or in the video, which I use chatgpt. Claude for the normal, you know, the asked questions, general questions or refine the.
You know, the writings under the as Josh mentioned. And also I use the perplexity the AI. To find the answer to do the search and also Google Google overview now is. A much better to give you a good summarization. And also I use AI search tools to help me to. Discover and help me to read the paper more efficiently. For example illustrates the search. Search space, et cetera and also the Gmail.
The AI writing stuff is very recently used. It's very. Very useful. I feel another category is about, you know, the AI ifo scholarly publishing. So in my experience. So there's a three most popular AI application. One is under the count basically for. Yeah so one is a content classification.
So how can we apply the AI to target the content to predict what topic this content talk about. So for example we have the article taxonomy and the auto tagging. We tag the automatically apply AI to tag the content, which is save about 1 minute per average per article. So imagine that how much we saved for on the millions of articles.
So another is recommendation. So this is another popular the AI application gives the content or based on the user's behavior we recommend the similar the content or the similar user and then speech recognition and give the article. This is also very important for Content Accessibility. So we automatically generate the captions. Or subscriptions or subscribe. And also the search discovery.
So now it's even more popular the content. Discovery because we can allow the user ask a natural language question get answer. So recently this is also the. So far, this is the most popular but recently also AFL integrity we already discussed. For example, the last year we have the release where we released the paper mill detection tools which apply AI to detect the, you know, the integrity issues, paper mill, the papers, et cetera.
So this is what the two categories in my mind. Yeah that's great. Thank you. How about you, rocio? I'll say that I don't go out of my way to go ask chatgpt anything. I often will just type it into Google and read the summary if I want to learn something quickly about something.
But as the other panelists have talked about, AI is everywhere already. Like it's in our outlook. It's in our Gmail. Like it will suggest what I want to say before I even type it in. So we are already using it in many aspects of what we do in our day to day. If we wanted to give a, you know, a use case, for example, especially if very good example, I think of how it's improving efficiencies.
We're also giving more opportunities for use if, let's say we had an image on a paper where we're concerned about manipulation or duplication or some other concerns. It used to be that five years ago we were having to have an image analyst sit-in with the image and spend maybe 30 minutes breaking it down and analyzing it in Photoshop. Now I can just group it into proofing or image twin, and within seconds I get a report that tells me what specifically are the issues with the image.
And that not only allows us to more efficiently look at things we're concerned about, but also just scan submissions, you know, regularly where we weren't even initially concerned about. And it's not only helping us catch fraudulent content, but also just innocent mistakes that happen. So it's improving the quality of the published output that we have across the industry.
So Yeah, that's very narrow. But I think effective example I'd have. Yeah Yeah. Is getting used. Absolutely that's great. So we did get a clarification in the Q&A. And it goes along with sort of what we're talking about now, but it's more like what's the standard. Do you think there will be a standard for how much is OK, for instance, how extensive the use can be for like grammar, how much can we use it for editing, for rewriting, like do you do you foresee a time when there's a, there's a standard that we come up with in publishing or like how much usage is OK?
I mean, we're sort of it's a conversation, but you know, do you think we'll. Do you think we'll get there, or do you think that's necessary? So come. Josh? Yeah. So I avoided this first question because I think it's a challenging question. Right like is there going to be one standard.
I think there will be standards, right. Just as there are standards of publishing. Right there's different peer review models. And I think that's good. Right we need to experiment. We need to test. We need to look at what are the efficiencies of access models versus subscription models versus I plus if you will, versus I negative if you will.
And so again, I think we need to kind of run all those tests as you know journals as publishers and understand and learn and share openly like what we have so we can get to some standards. Hey, here's what we did when we studied that. We improved reading of the articles by x. OK at what cost. Right and so I think without kind of having these analysis done, the standards are just going to be, you know, they're not evidence based glycolysis.
And so I think that's where we start, and I do think there will be some stronger standards, but I think it's a good thing that there's a diversity of journals and publishers and ways of doing what we do, because we'll continually evolve that model. Right and, you know, I find myself kind of like I used to. We need to change everything. And now I'm like, great that it hasn't changed.
And I think there's a mixture of both of those where it's a good thing we haven't changed everything about publications, right? They are permanent. They are archived. All these things are what make them so valuable and we should be resistant to just adopting everything and anything all at once. But we should try new things and experiment with that.
And so I don't know if that answers it. I do think there will be some standards, just as if there's like, you know, there's persistent identifiers, there's some standards for archival and peer review and things like that. And I think the groups, some of which we're all part of, you know, are going to help to, to lead to standardization, if you will, of that.
Yeah that's great. Having you here. Anything you'd like to add? Well, I'll just jump in to say that I agree with what Josh said. There are already working groups and discussions amongst like, you know, industry organizations on what are we going to agree on as a community of what is appropriate, what is not acceptable.
But it's really up to the publishers, I think, specifically to decide what is it specifically that they are willing to allow and not allow, and then just be transparent on what those policies are so that authors know how to prepare for those. And I mean, I can speak from the Aces policy. We allow transparent use, and then it's up to the editor to decide whether or not it's acceptable and whether or not it needs to be toned down or just outright rejected because it's too much use of AI.
But it really, I think it varies by field and also by what the scope of what the publisher is aiming for as they're thinking. And considering what their content is. Yeah, great. Anything to add? Yeah just add one more thing. I think the useful, as Josh said, this will be going to be the standard.
So I think there's no the unified the single standard. But the standard will be to keep updating I mean update. So until through the window, whenever there's a new scenario coming, the new technology happens. Even today, the human laws like the copyright laws is also under discussion. They need you know, update to for this especially for generative AI. So the standard for the publishing also I'm sure is going to be the updated.
So currently as you said it's we for the many publishers. We want to know the it's OK for to do to improve the grammar, leverage the AI to improve grammar, the writing style, et cetera but if you want to add you apply the AI to generate anything new, any other thing things sections or something you need to know that makes clear, transparent, and then it's for the journal editors to make the final decision, et cetera so be careful on when you use this.
Make sure you check the, you know, the guidelines. General guidelines. Yeah that's great. Yeah Thanks. As a follow on question from the chat related to transparency. Do you think disclaimers showing usage and amount of AI usage will become part of the standards? I think they're somewhat in many instructions for authors, but I'll Josh Hong we'll start with Josh then.
I think so, and I think right now it's kind of like a bad thing if you use it, right? It's something we're hiding. We're clicking that. We're not using it. And I think it's going to transition more to we are using it, here's how we're using it. And then it becomes more acceptable because of that transparency.
Right I think that transparency will help with the trust and the adoption of it. And so saying which tool you used it, why did you use it? I don't know, like all these things are again like interesting to kind of test. Is that more acceptable if you read an article like with these declarations versus if you didn't, I would think yes. Right you're being transparent.
You're being honest. Like this is where it is. And so I think transparency is a big one. I also think that goes to like publishers to vendors or tool builders and, you know, site like we are using AI and we sell to universities and we say, hey, we also publish a paper about how we built site, right, to give you the transparency into that block Black box. Because if you're going to teach it, if you're going to use it, like talking about the limitations, talking about the accuracy, all these things are part of, I think, the adoption of that.
And so that's a great question, Patricia. Transparency to me is like the key to adoption of it, because it's already been adopted, but just not in a transparent way. And I think if we want to get to a standard, transparency is crucial. I don't think we get there without that. So the for me, the transparency is also related to reproducibility because especially today, you know, when the more and more the researcher applies the generative AI with a relevant long prompts, different prompts to analyze all the data and generate results, et cetera.
So for in the disclaimers, maybe not only the tell them the which the AI tools they use, which version and how do they use? What's, the what's the data they are using and also the prompt. How do they ask what's the introduction, you know, to the AI, to how do they ask the AI to apply this. So with the prompt, so that maybe in the future whenever they need the, the, the, you know, the reviewers or editors, they can apply this to reproduce.
Also for other researchers, they can apply the prompt to reproduce the result in the future. So the transparency is Yeah, for me is not only about, you know, the an ability trust, but also reproducibility. Yeah, a good point. I'll agree with that. And also just echo that most policies on AI that I've seen, they do request that if an author has used AI to describe it in the methods section or wherever appropriate section, how it was used.
So there's already a disclaimer, so to speak, there, but it's aimed more at that. How do you make this reproducible aspect of it? Like how was it used? What are the parameters, what types, what versions of AI were used so that others can take what you've done and reproduce it? So Yeah. Great Thank you.
I wanted to get to at least one other of our questions, because I love to have our attendees be able to leave the talk with maybe a practical recommendation, but if one of our attendees has never used an AI tool, what's your recommendation? For one thing they can do or try or think about after they leave this webinar. So maybe.
Maybe I can jump to see something. Yeah Oh. Never use. I really don't know that anyone is never use. Eye to eye is everywhere. So focus on Google Search is as long as you use Google search. Amazon you go to Amazon. You all you already use. You know I the I behind.
So before but before using the I. So I want to emphasize something make some give some suggestion what for example before you use any AI tools you need to think about what's with this AI is designed to do is any limitations. So who owns your data. So this is a very important. We use the disclose any confidential information. This is a very, very be careful because sometimes we just really excited to use it without.
Consider this is your input being used for AI training. We need to make clear. For example in chatgpt there's a settings allow you to turn on turn off whether your data can be used for training. And have I done? Have you done any your own the accuracy check of it we cannot fully trust on I is not 100% accuracy, especially hallucination today.
So are you the prepared to take responsibility for this accuracy? You cannot just get grab the answers from AI and then use it to also does your institution. Does your company also use it as tools on your work? So all this I think they need to be careful to ask yourself before you use any AI tools. Absolutely Yeah. That's a great reminder.
Thank you so much Joshua. Rocio any from either of you? Yeah, I think so. That is a good word of caution. I'm glad you went first, because my suggestion is going to be just to try a bunch of them, right? Like I try the perplexity versus chatgpt. I would try a bunch of things that it's not sensitive information because a lot of them probably are eating that, but you get a good feeling.
And again, they're advancing. There's new models put out every month. And so even within one tool you can try a different model. And there's now these agentic approaches. There's different models that are they think or they take longer to answer. And so just kind of getting a feeling for it, even if you don't like it right, or want to use it. I think it's good to understand by just really testing and test on, like I always test on my PhD work and I'm like, OK, it's kind of good or bad.
That's my barometer. And so we're all like experts in something and we can kind of stress test them that way. And understand and be like, Oh wow, the citations really are bad in this, right? Or Oh wow, that's like, that's wrong. It was very quick, but it's not right. And I think getting that, you know, beyond all these studies that I keep saying we need to have is good to kind of understand and better evaluate.
And it's also maybe how you start to use them because after that you're like, Oh, wow, I was going to ask someone to do that, but now I can just ask this assistant to run this, like basic thing. So I test a lot of them and continually kind of test a lot. Yeah great. Did you have anything to add or. Oh, I think we lost your audio.
Sorry Oh, can you hear me? Oh, there we go. Now we're good. Now we're good. Yeah sorry. I was just going to echo what's already been said of I is here. We are going to be using it whether we want to or not. It's embedded into everything we do with technology nowadays. So how do we use it?
Responsibly and like Hahn said it perfectly. A lot of what you feed into an AI tool is used for training the tool and then probably producing further outputs. So as we're handling confidential information from like our own organizations, but also the content that we're considering if we're working in an office, if you put that into an AI bot to produce a review, how is that a then using it?
And how is that breaching confidentiality of the peer review process. So Yeah. Good point. Great we've been answering questions as we go, but there is one in the Q&A that's more of a moral question. How how would you address or reconcile the use of AI over human jobs? Then Kelsey is saying, for example, I started as a copy editor, but there's a high likelihood that copy editing as a profession won't exist within a few years.
Who does this advantage and disadvantage? Yeah and then Hannah followed up and just said, how do we navigate the inevitable integration of these AI tools in an ethical way? I just think that goes to training. The specific role might disappear, but new roles are going to be created. And how are companies going to train their workforce to adapt to these new technologies?
That's the way I look at it. I just added my $0.02, so I think the I will not know the AI is used to the arguments that humans work, facilitate human work, and then the. Whenever this new emerging technology coming, there's always some the jobs disappear, but the other new jobs coming for example the e-commerce when the internet is coming and then the many the physical shops closed down.
But the e-commerce in many the online shops opened. So this same today here. So as a result, I mentioned that some of the copy editor, maybe this new job, not only the Edit, the content for human, but they can edit, you know, the content for machine and using this all this content to help the machine AI to train them, you know, have the better training et cetera to maintain the metadata, et cetera. So this could be in the future.
I'm sure there's many the job will appear. This also, the can push the education, our education system we need. The education education system needs to be rethink the reorganized as well to reflect. This is my way. Yeah great. Yeah these are some challenging questions for Wednesday and very big questions because I think you can kind of think about like, Yeah, copy editing.
You can even think about like humanity, right. Like these companies are pushing not just to improve like the grammar of tools, but pushing for very, very big goals of like, you know, AGI, there's huge amounts of money behind it. And so, you know, this is scary in many ways. Like we don't know. And, you know, it's fueled by a lot of different motives and money.
And so there are so many ethical and moral and legal challenges that I think we need to kind of lean into it and understand that and push for things that we think are important. I do think Rocio gave a great answer. Training is important because we've long been impacted by technological changes. This is a big one, of course. And so there will be certain changes.
Some of those are going to be brutal and horrible, and some of those are going to be great and amazing. And so I think we got to kind of take it one step at a time. Otherwise you might just find the corner of the room and curl up. Yeah Yeah, exactly. Great Thank you. We have another question, and we may have sort of dressed it as we went along, but the question is, where is AI provided a business opportunity for your organization?
Anybody? Yeah. Hong Yeah. I think the AI is, you know, the we have discussed many the application exceeding expectations. So the AI in the scholar publishing especially for publisher, I can help publisher publish more, publish better and publish faster.
This is all about, you know, the opportunity about the reduced cost, increased revenue and also the share the more meaningful, useful information in the industry in the world. So this is already, it's a big opportunity. So Yeah. Great Well we're coming to the end here. So one maybe last question. Any books, websites or leaders in the area that you'd recommend our attendees sort of read follow to find out more information.
Any any sources that you love. So do you do you have a suggestion or I'll start with just say SSP Webinars are a great place. Yeah no, I don't know. I mean, it's I think a lot of just experiencing this and trying out tools is a great way. And then there's a lot of people. You know, making these big proclamations of what this is and what this isn't going to be.
I think going to the literature is great, right? We have a lot of answers. We are doing a webinar where it's like librarians looking at how chatgpt insight was used in the classroom. And it's interesting. Right? and so there's tons of different places you are yourself maybe publishing some of them on this call, which is fascinating.
And I think those are important because there's actual analysis and evidence behind them, like everyone's going to say we got the fastest, best tool. And we look at this benchmark. But how does that impact editorial? How does that impact publishing. And undoubtedly there's more and more studies starting to come out on that. So look within.
Yeah that's great. Anything else you'd like to add? Any hung or so. Yeah for me I usually the go to I found the Scholarly Kitchen is a good place to this. Although maybe because I'm also the chef of the one chef of the scholar Kitchen. So I introduced this and but this is really, you know, the thoughtful, you know, the things about, you know, the scholarly publishing there.
And also, as Josh mentioned, keep reading the research papers about the AI, the peer review. There's many things. And also, I usually go to the media that the, you know, the it's called the media. So it's very good, you know, the place to understand the many, you know, the deep complicated the, the concepts. And also I use AI tools like chatgpt perplexity to help me summarize, you know, what happened in the world.
Yeah great. Rocio, anything to add or. I think you lost your audio again. Sorry can you hear me now? Yes Yeah. Yeah great. I was just going to reiterate resources like SSP, the Scholarly Kitchen, but also just the CAC meetings.
Like any like big industry organizations are putting out really good content and that helps keep up with. Yeah, the changing landscape. And honestly, like I think Hong said it, but ask chatgpt to give you a summary of what's been in the news recently, and it's really good at doing that. That's great. Well, Thank you all.
This has been a really engaging discussion. I really appreciate your time today. And to our attendees, Thank you for your questions and for being here this morning. So you will receive an evaluation form by email within 24 hours. And a link to the evaluation form has been will be posted in the chat. And we encourage you to provide feedback to help us shape future programming.
And you can also visit the SSP website for information on future webinars. As you know, the discussion was recorded. All registrants will receive a link when posted on SSP website. You must be logged in with your SSP login credentials to view. So thank you all for joining us today and this concludes our session. Thank you.
Thank you. It's great.
Segment:0 .
Thank you, and welcome to today's SSP early career subcommittee webinar, from manuscripts to machines a fun dive into AI for publishing newbies.
We are very pleased that you could join us today. My name is Rebecca Benner and I'm a member of the SSP early career subcommittee. I have just a few housekeeping items to review before we get started. Attendee microphones have been muted automatically, so please use the Q&A box to enter questions for the panelists. We would love to hear your questions and have set aside time at the end to address them.
You can also use chat to say Hello, or to alert us to any technical problems you are having, and we encourage you to introduce yourself in the chat and tell us where you're from. Closed captions have been enabled. You can view captions by selecting the More option on your screen and choosing show captions. This one hour session will be recorded and made available to all in the next few days.
A quick note on SPS code of conduct and today's meeting. We are committed to diversity, equity and providing an inclusive meeting environment, fostering open dialogue free of harassment, discrimination and hostile conduct. We ask all participants, whether speaking or in chat, to consider and debate relevant viewpoints in an orderly, respectful and fair manner. It is now my pleasure to introduce our speakers today.
Josh Nicholson is the chief strategy officer of research solutions and co-founder of scite, a next generation citation index of citations into context. Previously, he was the founder and CEO of the winner, winner and CEO of authorea, two companies aimed at improving how scientists publish and collaborate. He holds a PhD in Cell biology from Virginia Tech, where his research focused on the effects of aneuploidy on chromosome segregation and cancer.
Rocio Vidal is publication data and ethics manager at the American Chemical Society. Rocio joined ACEs in June 2024 as a publication data and ethics manager. In her role, she provides guidance to ACEs journals on ethics cases, promotes best practices, and helps DraftKings policies in accordance with industry standards set by the Committee on publication ethics, for instance, and assesses research integrity tools for the ACEs publishing integrity office.
Prior to joining ACEs, Rocio worked as an editorial manager at science advances at ars. Hongzhou is the senior director of AI product management at Wiley, where he designs AI strategy, drives the product roadmap and leads teams in developing AI powered solutions that automate and enhance research and publishing processes.
Huang is a recognized a thought leader in the scholarly publishing industry. He articulates the vision, challenges, and opportunities of AI and transforming the field. Huang has been a Scholarly Kitchen chef since 2023, co-chairs alpsp, a special interest group, and also serves as a cope advisor and is a member of STEM future lab. Thank you Josh, Rocio and Hong for joining the panel today.
Before I turn to our panel discussion, we have a poll we would love for you all to complete. It should come up on your screen shortly and we would love to know. Do you use AI in your professional lives and work? And do you think you are using AI more or less than your colleagues?
It looks like most have answered. All right. Great Well, Thanks for Thanks for sharing your experience with AI. It's great to have a sense of how you all feel and what your experience looks like. So as we get started, feel free to share in the chat your name and location.
If you have questions for the panelists. Please enter them in the Q&A. All right, so my first question for the our panelists is, could you describe your pathway into publishing and how it led to your involvement with AI tools or ethical issues related to ai? I'll start with you, Josh. Yeah, Thanks, Rebecca. Hi, everyone.
It's nice to see the location. I thought that was fun. And so as you introduce me, my background is not in AI. My background is in Cell biology. As a PhD student, and even before then, I've always been interested in how science works. And so how do we publish, you know, how do we fund science, etc.? And so I used to just write these kind of pieces as a student that I would publish, analyzing citations, analyzing how we fund things slowly wanted to go beyond just publishing and to try and build some of these solutions.
And so the winnower was my first Foray into basically trying to fix everything I saw wrong with publishing. And so very naive and idealistic, you know, PhD student, not completely crazy though, so we started to publish things first and then winnow the good from the bad via post-publication peer review. So we started at the same time as like biorxiv ahead of my arxiv.
We were mentioned in all those different things. And it's funny, because if I look back on my emails from like 2013, I have emails saying, hey, we should start a archive. I spelled it differently than the current one. And so that's kind of how I got into, I would say, you know, publishing, if you will, was trying to launch an alternative, you know, publishing Foray from that.
You know, the winnower was not just aimed at trying to like be a new publisher, but was really trying to address the challenges of reproducibility. And so that is where site fits in. And, you know, it's funny because on the winnower, we published a paper suggesting this new way of looking at citations. Right wouldn't it be great if you could see if a citation was not just, you know, 1 to 1, but exactly the type of citation, how and why and where that citation came from.
And so started very manually analyzing hundreds of papers and looking at sentences where the citations were made realized you could break down and bucket citations into different types. And of course, there's literature looking at citation ontologies, but realized also, if you want to do this at scale, you need AI, right? You need some solution, right? Because we're talking about billions of sentences from research articles.
And so we I've been trying to talk to anyone that would listen to me to say, hey, you got this idea. Can we automate this for a long time? And I think we started building website at the right time because there was enough advances in AI to do what we do, which is classifying a citation statement as supporting or contrasting. And to do that, you know, with some level of accuracy that is acceptable.
And so that's where we are. You know, we spent a lot of time partnering with publishers, both publishers actually on the call to help build out these next generation citations and display them on the version of record. And I would say that is, you know, and this will maybe get to something we speak later. Kind of the more vanilla view of AI, right? We do use machine learning and deep learning to do this.
But, you know, historically that was not anything to do with generative AI. So that's kind of how I got into it, where we are with AI, and also why we started to use AI, not just because it was there, but to solve a problem. That's great. Thank you so much. Rocio, how about you?
All right. So I have a pretty different, I guess, upbringing through my career. But basically I did also start in a graduate degree. I very quickly into the graduate program decided that I didn't want to be a researcher myself, but I did want to contribute in some capacity to the scientific community. So I found myself applying to anything science related.
I ended up getting an entry level editorial coordinator job at a journal where I learned everything from, you know, submission through publication, like handling the peer review. Working in editorial journal. Operations, et cetera. And I agree with you that role into management of editorial staffs. And then, you know, early to 2021, 2020, when we were starting to see the uprise of the distrust in science through, you know, the pandemic, but also the growth of paper Mills, I started to get more and more interested in these ethical questions and dilemmas that we were facing as an industry.
And what better way than to learn more than just become like, have that become part of my role? So I decided to take on this new role as a data and ethics manager. And I've been, Yeah, learning a lot and growing a lot in the ethics space, especially when it comes to AI, but also other issues that we're facing as an industry. Yeah that's great.
Wonderful and Hong, how about you? Right I'm also different from different path. So after my working as a software engineer, lead and MBA. So I joined the Wiley and as a senior product manager for the information discovered in the 2017. So right when the deepmind the alphago made the history. So at that time. So with my PhD, because my PhD is in the 3D modeling with AI and the MBAs in the digital transformation.
So I knew I will be the future or the future. So I started building the why join the company and start building the AI teams, including the R&D and engineering teams. And also at that time, because the most of the customer people don't have the, you know, the confidence for the AI and it's still early. So we have to, you know, the identify the every possible the opportunities, build the prototype solutions to prove the value.
It's not easy. But you know, we're making progress. So and also that I'm very happy and feel the I'm glad in the I work in the scholarly publishing industry because this is a unique place industry for industrial Ii because our content, you know, the few, the millions, millions of content fields I training and also the I intern the enhance everything from authoring peer review to publish and discovery.
So this is a unique place to work on the AI. Yeah Yeah. So that's great. I think that leads really well into our next question, because I think it might be helpful for our attendees to sort of be clear about what we're talking about when we're talking about AI. And so maybe I'll start with you. Could you provide some context related?
You know, you were just mentioning, you know, our day to day activities and generative AI. So could you sort of level set for the our attendees? Yeah, I think for me it's a 2023. You know, two years ago 2023 is a clear the milestone before 2023. AI in publishing was mostly a task specific driven rebuilding the machine learning model. Deep learning model.
As Josh mentioned for a specific task, we built this for the recommendation for classification, for speech recognition, et cetera but after the 2023 is end of the, because it is the end of December of 2022. So Thanks to GPT and the large language model. So we have shifted from to a more unified approach. I mean, now a single large language model can handle translation, summarization or text generation and even complex Q&A tasks.
Question answered, you know, task which which is a very hard problem, which is almost impossible for the, you know, the individual, you know, the machine learning or deep learning model to do. But now it's possible. So the entry barrier technology barrier has become lower. So it makes you know, the product and the platform which integrates this AI technologies even more important already. That's great.
Josh, would you like to add to that? Yeah I mean, I would really just echo that. I think you captured it very well, right? We have been using AI, you know, for some time. And you can find Eugene Garfield talking about AI in the 60s, right? And so AI has been around. I don't think it's been appreciated because it was very task specific.
It wasn't so easily accessible. And I really think this 2023 was a defining year where chatgpt came out. You know, I think it's the fastest growing product ever, and we can all appreciate it. And I think that's why we're all talking about AI nowadays, right? Because AI has been around. We've been talking about it.
Theoretically, people are doing PhDs on it. But here is this kind of transformative moment that, you know, is just this powerful tool, very easily accessible, that you can plug into different systems and tools you know, quite easily and experience it. Right before you had to kind of be an engineer, understand this, and now you can go to chatgpt or any of these things. And you feel kind of that difference there. And so it's really kind of this, this big inflection point.
And I think it's interesting. But I do think it's important to kind of consider both. Right because it's so new we can be so scared of it. But there's also opportunities. And we have been using variations of AI for quite some time. Great Thank you. Rocio, how about you? Would you like to add anything?
No I mean, I would just continue to echo what it's already been said that really generative AI has brought up a disruption to the industry and all industries, really, and that we now have to consider things that we haven't even considered from a philosophical or an ethical point of view. Like what? How do we use it? Does it replace a review or does it replace an editor? How does our role change now that we have access to these tools, and how do we do that responsibly?
So Yeah, it's a very interesting time. Yeah, absolutely. That's great. So this is for everybody. But could you share your perspective on what types of AI are having the biggest impact that you think on scholarly publishing and what those impacts are? And maybe I'll start with you, Rocio, this time. Sure so, I mean, I already touched on my previous answer, but thinking about redefining what our role is as we use this tool, where are the guardrails for using it?
How do we define the use? Why do we, as an industry decide it's acceptable versus unacceptable use? And what impact is that going to have on how science evolves moving forward? I mean, when you think about AI and I'm sorry, science and reproducibility it like AI right now, at least to me, I'm not an expert on it, but it's very much a Black box where you feed something into a large language model and it spits something else out.
What's actually happening in the background? How do we build scientific content that's reproducible? So that is, I guess, questions that are going to have to be answered. And you know, as an industry, we're moving in the right direction. But still there's a lot to discuss and to define. Yeah that's great. Thank you.
Josh, how about you? Yeah, I think Rocio has got a tough job because ethical challenges, legal challenges to me are kind of like the biggest impact, if you will. Right? the technology is so far ahead of everything that people are using it. We don't know what to do with a policy whether, you know, we're at a journal or a University, and then again two weeks later, like there's some bigger advance, right?
And so the technology is moving at such a rapid rate that it's very hard to do anything from a legal or ethical or like pedagogical perspective, because it just changes so dramatically. And I do think that's where it's going to be maybe the most critical. Right can we use this for authoring? Right people are using it already. Is it enough to put a checkbox and say don't use it?
Well, as we see, like it's pervasive, they're going to use it for peer reviewing. We're telling them not to. But again, we need to kind of figure out the ethical, you know, applications and policies that we as publishers, we as teachers, we as researchers have because it's here. Right and people will use it. It does save time.
I mean, it's pretty amazing for certain things. But we need to recognize the risks, the challenges, the legal things. And so, you know, also what Hong said, I think it's a privilege to kind of be in our space of, of research, publishing, because LMS have ate up the entire part of the web, if you will. They haven't really touched research articles in any way.
And I think research articles, if you go back to the fundamentals, like are very valuable. And I don't mean that from just a monetary perspective, but from the content they have, right? They touch every aspect of our life. There's research on how Peppa Pig influences your children learning English as a second language to prostate cancer, to particle physics. They are backed by evidence, statistics, peer review.
And so how do we like how do we leverage these things together. Right this trust. And I think scholarly publications and even citations which I'm biased, obviously by that have a role here. And it's a very important role because LMS are hard to trust. And so, Yeah, it's interesting. But to me, I think the biggest impacts are going to be like some of these legal challenges.
And then also, how do we responsibly enter this, this post 2023 chatgpt world. Yeah, absolutely. I think, you know, those are some of the conversations we're having at our organization. You know, how it you know, it feels like there's some tasks that really that could utilize AI and really save some people time. But then there's we definitely keeping the legal and other aspects at the front of mind too.
So it's sort of being careful about that is definitely something we've been doing for sure. How about you, hon? What would you like to add? Right I think I try to, you know, the talk of something, some of the type of I from the mainly from the technology and the solution application perspective. So for me, there's several, the type of eyes I believe that have the biggest impact has already happened.
For example, the multi-modal the large language model, the AI which the like the GPT four. So it's not only the handle the generate the tax understand generate tax but also the handle the images, audio, video et cetera seamlessly. What does it mean to the research? Research means we can help the researcher to generate the much richer, you know, the input, research input and output.
So in the future it's not only the paper, not only maybe text based paper, but maybe, you know, the audio, video, and the corresponding images and the kind of stuff and Ar VR in the future, of course. And another is the right pipeline retrieval, augmentation. The generative pipeline, which is almost is a foundation for almost all the, you know, the discovery system today.
I mean, the generative AI based, the discovery system today is a conversational discovery. So to make sure that people not search for the keyword by keywords, search for keywords, return 100 results. You still need to find answer by themselves, but you can like talk to the experts and then the iteratively and until you find the answer and they get answer directly from the, you know, the millions of, the documents. So I think this is and also in the future I, I assume you know, the most of the content on the internet will be the AI generated in the future.
And another is the AI agent. So this will be very popular where everyone talk about. So this AI system which is at autonomously the break down the tasks, search for data, for example, the search for data. Use the tools intelligent so user can just specify their goals with agents. They can automatically break down the goal into the different tasks and decide what and where to search to, what to do, which tool to use.
So this is very, very useful. And and another thing is not about AI application tools itself, but about AI governance platform. So because as AI grows and the risk also grows of course. So they have these tools we need. We need to know the specific dedicate the tools, many governance tools to help us organize and manage the risk, comply with the regulations because the different, you know, the maybe different the country, the area have the different regulation for AI.
We need to make sure the align the our, the AI development, deployment with the best AI practice with ethical standards, et cetera by embedding the governance mechanism through the whole the AI life cycle, not just the AWARE when you use it, but to start when you have this idea, when you start building this and when deploy it or you, you just try to, you know, buy from the third party vendor or something. So all these needs of aligning with, you know, the ethical standard.
Yeah that's great. Good considerations. So let's move on and talk a little bit about opportunities and threats. I mean I think it's sort of a thread through all, you know, everything we've been talking about so far. But you know, as a specific question. Could you talk about some of the opportunities and, and threats that are attendees should be or aware of now or maybe be thinking about for the future?
So I'll start with you, Josh. Yeah I. Where should I start? So I think in terms of opportunities, I think there's a lot of uncertainty. Right and in uncertainty there's opportunities as well. And so to me I do think AI is not going away. And so we need to kind of come to the table as like what does that mean if it's not going away. How is this going to be used in peer review.
Not is it going to be used in peer review? And you know, from the perspective of starting companies, I do think there's interesting things to test where it's like, maybe this is not going to be the standard right now, but we can set aside, run some experiments, test different models, whether that's peer review AI systems, whether whatever it is to test, get numbers, look at this and understand here's a world with AI natively.
Right and I think there's opportunities for people to do that, you know, by starting companies to test individually. Because I do think New workflows are going to emerge out of this. Not all the workflows that we might love, which is maybe a threat, but there's going to be changes and we have to kind of come to the table with a sober view of, here is this powerful tool. It's going to keep getting more powerful.
What does that mean? Right that improve some of my day to day workflows? You know, it improves my emails. I draft an email and I'll say fix the grammar. Like make this more convincing. And it's what I had written. But then it like removes all the grammatical errors and the weird, you know, things that I had. And so I think it's, you know, we all have different kind of perspectives.
We're all doing slightly different things, probably in our day to day. And thinking about it from an n of one perspective can help, but then also thinking about it from like a perspective of like the industry. Is there opportunities that I see that make sense, that I think will become mainstream in the future? And going back to like my first introduction, like preprints weren't a thing in biorxiv, right?
You know, or in biology. Right they became a thing. Right? and I think starting to think of that from the AI perspective is what is going to be kind of the archive of I, if you will. Right and is that an opportunity for a publisher? For an editor? And I, you know, as a scientist, there's this saying, and I won't say it here because there's some curse words in it, but it's like, do the experiment.
Right test it. Right go test these things because everyone's going to have different opinions and thoughts. But ultimately, and this is why research is so amazing and why, like I am in this space, we need to test these things. How does this impact pedagogy? How does this impact peer review. How does this impact writing. And from that, we can make better policy, ethical decisions, et cetera.
But until then it's a lot of different shouting and me waving my hands even more than I'm waving them now. Yeah, I think that's a really helpful way to think about it. Approach it. You know, that's it's almost in a way, you know, like how we approach anything else. It just feels very large and impossible, maybe. But, you know, if we approach it using the same methods we've maybe used for other things, we can break it down and figure it out.
Maybe Yeah. How would you like to add? As you think about opportunities and threats that our attendees might want to know about. Sure to build on what Josh just mentioned, I think the AI is definitely is a very useful tool to increase the, you know, the productivity in our daily work, you know, work life or life. I mean, the for example, I think maybe the many people are already used, you know, the Google, you know, the Gmail, you can they provide based on the Google Gemini, the model, they can automatically draft help you to draft the email or something.
And also in the teams meeting Microsoft Teams meeting before we can just we can either accept or decline the meeting event. But now we can follow. We can follow the meeting. So there's a third options. So even if you don't attend the meeting they will still send you the record and the transcript automatically generated by AI and even the summarization.
You know the summary and also action point, which is very useful. So this is already the changed the way of our work. And from the research perspective, for researcher itself, I think the I can be used will be in the future. You know, the imaging of ais and not only the fund organized information, but also the think and the Apply the knowledge and help them plan a executed, you know, the research to work faster.
So it's basically it's like the personal the research assistant. This is, today in the Google. Apple already have the general assistant. But in few future we should have the is happening now. So we should every researcher should have a research assistant to help you to speed up the research work. So this is another thing another two opportunities. One is the AI for peer review.
Because one of the top challenges in the scholarly publishing is a slow, as you know, the slowdown, slow of the, you know, the peer review process. And then how can we apply AI to speed up this? So I can help you find the relevant reviewers from the, you know, the many, the 100, thousands of the, you know, the revered author database. And can we even generate the initial report review report based on the readability, novelty, reproducibility?
This is also happen. We can see, there are many more and more. The you know, the startup and the company is coming into to touch this area. And another area is AI for integrity. This is a bit special. This because it has both opportunity and the threat. So opportunities we can apply AI to better detect this the fraudulent papers, the paper mill papers is the AI generated content is in the relevant scope, in the scope of journal, et cetera but also the well the AI boost, the search quality.
It's also the integrity concern itself. So plagarism image manipulation paper mill. So which all this makes it much easier to generate all this. You know, the fake papers, fake images by using AI. So this so that makes detecting the AI generated fraudulent will be even harder. So this is an ongoing battle. Yeah, absolutely. And that's a great segue to Rocio.
You could talk about your perspective working in ethics. And what you see as opportunities and threats. Well, I'll start with a positive side on the opportunities. I think AI often opens a lot of doors for access, both for the authors that are, you know, maybe using it to improve their language, to improve their writing, to make it more convincing, like there is already data to show that a quality, high quality written paper improves your chances of getting accepted and published.
So I think it's going to open a lot of doors for non-native speakers and people that just struggled with the writing aspects of scientific work to contribute to the conversation in terms of, you know. Lessening the workload of reviewers. We also are already seeing tools that are helping identify new reviewers that an editor might have not even considered, or potentially like find ways where an I can focus on whether or not the paper is reproducible and that the reviewer, the peer expert, talk about potential ways to improve this research or to like, you know, find new directions to like, take the paper forward.
But then also on kind of like the, the access to the like content that we're publishing, AI offers a great opportunity to create plain language models that, like patients, for example, can use as they're like considering treatment methods or like generally people just are interested in the scientific content that are currently, you know, there is a language barrier, scientists write, for scientists, and the general public can't really access that very readily.
So I just opens that door to a broader audience. On the risk side, I mean, I wouldn't work in scientific publishing or, sorry, industry. The integrity side of it without talking about fraud. And like Hong said, it is AI is giving a very helpful tool to fraudulent content and people that want to create fraudulent content. So when we used to think about fraud five, 10 years ago, we were thinking about plagiarism, about someone opening Photoshop and drawing on their images or like beautifying them.
Now, why would you go and plagiarize someone when it's easier to just ask an AI bot to create an entirely new paper for you that gives you new text, new images, new video content? And like Han said, it is very difficult to police this and catch these as we and our vendors are like developing tools to catch and detect fraud. The paper Mills are also developing tools to circumvent those tools that we have access to.
So it is an ever evolving landscape. And, you know, we're there's very smart people working to create tools and to, you know, make advancements in that space. But it's also a growing and continuing challenge there. So we'll see what comes out of it. Yeah, absolutely. Before we move to our next question, there is a question from one of our attendees.
And I think it probably fits well with what we're talking about. But would the publishing industry move towards a standard i.e. usage practice, and if so, what can we expect? Interesting question. And I would like to answer what can you elaborate the more what's the standard i.e. usage practice?
What do you refer to? So maybe maybe he can clarify or they can clarify in the Q&A or in the chat. But we can also come back to it. So maybe if you could clarify that in the chat, we'll come back to your question. But for the next one then could you all talk about how you use AI in your daily activities.
So more from a practical standpoint, you know, a use case or, you know, it can be something as broad as that or just really just nitty gritty. How do you use it daily. And I'll start with you Josh. Yeah so I mentioned one kind of thing that I do, which is just cleaning up emails and trying to do some like data analysis with it.
And so I'll instead of like using Excel I guess, or like going to grammarly, I'll just copy and paste and say, here's my draft that I wrote, fix it. Like make it more convincing, make it nicer, whatever I want. And so I do use it in that way, and I think it's pretty powerful, right? It like, makes my email a lot more easy to read right to Rocio's point, it's helping, right? I am an English like native English speaker, but I don't have the best grammar.
And so that, you know, improves that. And I think I do use it kind of day to day looking at different data analysis, just like kind of back of the napkin math that I used to do. Now I'll have chatgpt kind of do for me, you know, beyond that, like we are seitai before LMS exist, right? So we are always using AI for a variety of different things. And I do think, you know, we're mostly talking about LMS here, but, you know, we use AI to classify citations.
This gets to integrity, reproducibility and looking at all these things I think it also helps with discoverability as well. And so we're constantly working with publishers to think about, you know, how can this help from kind of the product perspective, but then also the licensing perspective. And so scite joined research solutions, which was formerly preprints desk about 14 months ago.
Research solutions is doctor'll write delivering documents, you know, supplying the access and rights to articles on an a La Carte basis to large corporations. We think a lot about how can we partner with publishers to get some clarification on AI rights. Right is there a mechanism that allow you, allows you, as a publisher, to bridge this gap from like OpenAI to publisher, where there can be some use, right. There can be some control, we can answer some of these tough ethical, legal, monetary perspective questions.
I wouldn't say we've solved all of that, but I think, you know, we are partners to publishers. We're independent from publishers. And we think about it from the product perspective, the editorial perspective, and then the licensing and rights perspective as well. And so constantly thinking about it and then, Yeah, you know, just using it kind of maybe in the way that everyone kind of uses it to help with little tasks here and there.
Yeah Yeah. That's great. Thanks, Josh. How about you, hon? Right so I think that we can. I can try to put it into two categories. I for the daily work for myself or in the video, which I use chatgpt. Claude for the normal, you know, the asked questions, general questions or refine the.
You know, the writings under the as Josh mentioned. And also I use the perplexity the AI. To find the answer to do the search and also Google Google overview now is. A much better to give you a good summarization. And also I use AI search tools to help me to. Discover and help me to read the paper more efficiently. For example illustrates the search. Search space, et cetera and also the Gmail.
The AI writing stuff is very recently used. It's very. Very useful. I feel another category is about, you know, the AI ifo scholarly publishing. So in my experience. So there's a three most popular AI application. One is under the count basically for. Yeah so one is a content classification.
So how can we apply the AI to target the content to predict what topic this content talk about. So for example we have the article taxonomy and the auto tagging. We tag the automatically apply AI to tag the content, which is save about 1 minute per average per article. So imagine that how much we saved for on the millions of articles.
So another is recommendation. So this is another popular the AI application gives the content or based on the user's behavior we recommend the similar the content or the similar user and then speech recognition and give the article. This is also very important for Content Accessibility. So we automatically generate the captions. Or subscriptions or subscribe. And also the search discovery.
So now it's even more popular the content. Discovery because we can allow the user ask a natural language question get answer. So recently this is also the. So far, this is the most popular but recently also AFL integrity we already discussed. For example, the last year we have the release where we released the paper mill detection tools which apply AI to detect the, you know, the integrity issues, paper mill, the papers, et cetera.
So this is what the two categories in my mind. Yeah that's great. Thank you. How about you, rocio? I'll say that I don't go out of my way to go ask chatgpt anything. I often will just type it into Google and read the summary if I want to learn something quickly about something.
But as the other panelists have talked about, AI is everywhere already. Like it's in our outlook. It's in our Gmail. Like it will suggest what I want to say before I even type it in. So we are already using it in many aspects of what we do in our day to day. If we wanted to give a, you know, a use case, for example, especially if very good example, I think of how it's improving efficiencies.
We're also giving more opportunities for use if, let's say we had an image on a paper where we're concerned about manipulation or duplication or some other concerns. It used to be that five years ago we were having to have an image analyst sit-in with the image and spend maybe 30 minutes breaking it down and analyzing it in Photoshop. Now I can just group it into proofing or image twin, and within seconds I get a report that tells me what specifically are the issues with the image.
And that not only allows us to more efficiently look at things we're concerned about, but also just scan submissions, you know, regularly where we weren't even initially concerned about. And it's not only helping us catch fraudulent content, but also just innocent mistakes that happen. So it's improving the quality of the published output that we have across the industry.
So Yeah, that's very narrow. But I think effective example I'd have. Yeah Yeah. Is getting used. Absolutely that's great. So we did get a clarification in the Q&A. And it goes along with sort of what we're talking about now, but it's more like what's the standard. Do you think there will be a standard for how much is OK, for instance, how extensive the use can be for like grammar, how much can we use it for editing, for rewriting, like do you do you foresee a time when there's a, there's a standard that we come up with in publishing or like how much usage is OK?
I mean, we're sort of it's a conversation, but you know, do you think we'll. Do you think we'll get there, or do you think that's necessary? So come. Josh? Yeah. So I avoided this first question because I think it's a challenging question. Right like is there going to be one standard.
I think there will be standards, right. Just as there are standards of publishing. Right there's different peer review models. And I think that's good. Right we need to experiment. We need to test. We need to look at what are the efficiencies of access models versus subscription models versus I plus if you will, versus I negative if you will.
And so again, I think we need to kind of run all those tests as you know journals as publishers and understand and learn and share openly like what we have so we can get to some standards. Hey, here's what we did when we studied that. We improved reading of the articles by x. OK at what cost. Right and so I think without kind of having these analysis done, the standards are just going to be, you know, they're not evidence based glycolysis.
And so I think that's where we start, and I do think there will be some stronger standards, but I think it's a good thing that there's a diversity of journals and publishers and ways of doing what we do, because we'll continually evolve that model. Right and, you know, I find myself kind of like I used to. We need to change everything. And now I'm like, great that it hasn't changed.
And I think there's a mixture of both of those where it's a good thing we haven't changed everything about publications, right? They are permanent. They are archived. All these things are what make them so valuable and we should be resistant to just adopting everything and anything all at once. But we should try new things and experiment with that.
And so I don't know if that answers it. I do think there will be some standards, just as if there's like, you know, there's persistent identifiers, there's some standards for archival and peer review and things like that. And I think the groups, some of which we're all part of, you know, are going to help to, to lead to standardization, if you will, of that.
Yeah that's great. Having you here. Anything you'd like to add? Well, I'll just jump in to say that I agree with what Josh said. There are already working groups and discussions amongst like, you know, industry organizations on what are we going to agree on as a community of what is appropriate, what is not acceptable.
But it's really up to the publishers, I think, specifically to decide what is it specifically that they are willing to allow and not allow, and then just be transparent on what those policies are so that authors know how to prepare for those. And I mean, I can speak from the Aces policy. We allow transparent use, and then it's up to the editor to decide whether or not it's acceptable and whether or not it needs to be toned down or just outright rejected because it's too much use of AI.
But it really, I think it varies by field and also by what the scope of what the publisher is aiming for as they're thinking. And considering what their content is. Yeah, great. Anything to add? Yeah just add one more thing. I think the useful, as Josh said, this will be going to be the standard.
So I think there's no the unified the single standard. But the standard will be to keep updating I mean update. So until through the window, whenever there's a new scenario coming, the new technology happens. Even today, the human laws like the copyright laws is also under discussion. They need you know, update to for this especially for generative AI. So the standard for the publishing also I'm sure is going to be the updated.
So currently as you said it's we for the many publishers. We want to know the it's OK for to do to improve the grammar, leverage the AI to improve grammar, the writing style, et cetera but if you want to add you apply the AI to generate anything new, any other thing things sections or something you need to know that makes clear, transparent, and then it's for the journal editors to make the final decision, et cetera so be careful on when you use this.
Make sure you check the, you know, the guidelines. General guidelines. Yeah that's great. Yeah Thanks. As a follow on question from the chat related to transparency. Do you think disclaimers showing usage and amount of AI usage will become part of the standards? I think they're somewhat in many instructions for authors, but I'll Josh Hong we'll start with Josh then.
I think so, and I think right now it's kind of like a bad thing if you use it, right? It's something we're hiding. We're clicking that. We're not using it. And I think it's going to transition more to we are using it, here's how we're using it. And then it becomes more acceptable because of that transparency.
Right I think that transparency will help with the trust and the adoption of it. And so saying which tool you used it, why did you use it? I don't know, like all these things are again like interesting to kind of test. Is that more acceptable if you read an article like with these declarations versus if you didn't, I would think yes. Right you're being transparent.
You're being honest. Like this is where it is. And so I think transparency is a big one. I also think that goes to like publishers to vendors or tool builders and, you know, site like we are using AI and we sell to universities and we say, hey, we also publish a paper about how we built site, right, to give you the transparency into that block Black box. Because if you're going to teach it, if you're going to use it, like talking about the limitations, talking about the accuracy, all these things are part of, I think, the adoption of that.
And so that's a great question, Patricia. Transparency to me is like the key to adoption of it, because it's already been adopted, but just not in a transparent way. And I think if we want to get to a standard, transparency is crucial. I don't think we get there without that. So the for me, the transparency is also related to reproducibility because especially today, you know, when the more and more the researcher applies the generative AI with a relevant long prompts, different prompts to analyze all the data and generate results, et cetera.
So for in the disclaimers, maybe not only the tell them the which the AI tools they use, which version and how do they use? What's, the what's the data they are using and also the prompt. How do they ask what's the introduction, you know, to the AI, to how do they ask the AI to apply this. So with the prompt, so that maybe in the future whenever they need the, the, the, you know, the reviewers or editors, they can apply this to reproduce.
Also for other researchers, they can apply the prompt to reproduce the result in the future. So the transparency is Yeah, for me is not only about, you know, the an ability trust, but also reproducibility. Yeah, a good point. I'll agree with that. And also just echo that most policies on AI that I've seen, they do request that if an author has used AI to describe it in the methods section or wherever appropriate section, how it was used.
So there's already a disclaimer, so to speak, there, but it's aimed more at that. How do you make this reproducible aspect of it? Like how was it used? What are the parameters, what types, what versions of AI were used so that others can take what you've done and reproduce it? So Yeah. Great Thank you.
I wanted to get to at least one other of our questions, because I love to have our attendees be able to leave the talk with maybe a practical recommendation, but if one of our attendees has never used an AI tool, what's your recommendation? For one thing they can do or try or think about after they leave this webinar. So maybe.
Maybe I can jump to see something. Yeah Oh. Never use. I really don't know that anyone is never use. Eye to eye is everywhere. So focus on Google Search is as long as you use Google search. Amazon you go to Amazon. You all you already use. You know I the I behind.
So before but before using the I. So I want to emphasize something make some give some suggestion what for example before you use any AI tools you need to think about what's with this AI is designed to do is any limitations. So who owns your data. So this is a very important. We use the disclose any confidential information. This is a very, very be careful because sometimes we just really excited to use it without.
Consider this is your input being used for AI training. We need to make clear. For example in chatgpt there's a settings allow you to turn on turn off whether your data can be used for training. And have I done? Have you done any your own the accuracy check of it we cannot fully trust on I is not 100% accuracy, especially hallucination today.
So are you the prepared to take responsibility for this accuracy? You cannot just get grab the answers from AI and then use it to also does your institution. Does your company also use it as tools on your work? So all this I think they need to be careful to ask yourself before you use any AI tools. Absolutely Yeah. That's a great reminder.
Thank you so much Joshua. Rocio any from either of you? Yeah, I think so. That is a good word of caution. I'm glad you went first, because my suggestion is going to be just to try a bunch of them, right? Like I try the perplexity versus chatgpt. I would try a bunch of things that it's not sensitive information because a lot of them probably are eating that, but you get a good feeling.
And again, they're advancing. There's new models put out every month. And so even within one tool you can try a different model. And there's now these agentic approaches. There's different models that are they think or they take longer to answer. And so just kind of getting a feeling for it, even if you don't like it right, or want to use it. I think it's good to understand by just really testing and test on, like I always test on my PhD work and I'm like, OK, it's kind of good or bad.
That's my barometer. And so we're all like experts in something and we can kind of stress test them that way. And understand and be like, Oh wow, the citations really are bad in this, right? Or Oh wow, that's like, that's wrong. It was very quick, but it's not right. And I think getting that, you know, beyond all these studies that I keep saying we need to have is good to kind of understand and better evaluate.
And it's also maybe how you start to use them because after that you're like, Oh, wow, I was going to ask someone to do that, but now I can just ask this assistant to run this, like basic thing. So I test a lot of them and continually kind of test a lot. Yeah great. Did you have anything to add or. Oh, I think we lost your audio.
Sorry Oh, can you hear me? Oh, there we go. Now we're good. Now we're good. Yeah sorry. I was just going to echo what's already been said of I is here. We are going to be using it whether we want to or not. It's embedded into everything we do with technology nowadays. So how do we use it?
Responsibly and like Hahn said it perfectly. A lot of what you feed into an AI tool is used for training the tool and then probably producing further outputs. So as we're handling confidential information from like our own organizations, but also the content that we're considering if we're working in an office, if you put that into an AI bot to produce a review, how is that a then using it?
And how is that breaching confidentiality of the peer review process. So Yeah. Good point. Great we've been answering questions as we go, but there is one in the Q&A that's more of a moral question. How how would you address or reconcile the use of AI over human jobs? Then Kelsey is saying, for example, I started as a copy editor, but there's a high likelihood that copy editing as a profession won't exist within a few years.
Who does this advantage and disadvantage? Yeah and then Hannah followed up and just said, how do we navigate the inevitable integration of these AI tools in an ethical way? I just think that goes to training. The specific role might disappear, but new roles are going to be created. And how are companies going to train their workforce to adapt to these new technologies?
That's the way I look at it. I just added my $0.02, so I think the I will not know the AI is used to the arguments that humans work, facilitate human work, and then the. Whenever this new emerging technology coming, there's always some the jobs disappear, but the other new jobs coming for example the e-commerce when the internet is coming and then the many the physical shops closed down.
But the e-commerce in many the online shops opened. So this same today here. So as a result, I mentioned that some of the copy editor, maybe this new job, not only the Edit, the content for human, but they can edit, you know, the content for machine and using this all this content to help the machine AI to train them, you know, have the better training et cetera to maintain the metadata, et cetera. So this could be in the future.
I'm sure there's many the job will appear. This also, the can push the education, our education system we need. The education education system needs to be rethink the reorganized as well to reflect. This is my way. Yeah great. Yeah these are some challenging questions for Wednesday and very big questions because I think you can kind of think about like, Yeah, copy editing.
You can even think about like humanity, right. Like these companies are pushing not just to improve like the grammar of tools, but pushing for very, very big goals of like, you know, AGI, there's huge amounts of money behind it. And so, you know, this is scary in many ways. Like we don't know. And, you know, it's fueled by a lot of different motives and money.
And so there are so many ethical and moral and legal challenges that I think we need to kind of lean into it and understand that and push for things that we think are important. I do think Rocio gave a great answer. Training is important because we've long been impacted by technological changes. This is a big one, of course. And so there will be certain changes.
Some of those are going to be brutal and horrible, and some of those are going to be great and amazing. And so I think we got to kind of take it one step at a time. Otherwise you might just find the corner of the room and curl up. Yeah Yeah, exactly. Great Thank you. We have another question, and we may have sort of dressed it as we went along, but the question is, where is AI provided a business opportunity for your organization?
Anybody? Yeah. Hong Yeah. I think the AI is, you know, the we have discussed many the application exceeding expectations. So the AI in the scholar publishing especially for publisher, I can help publisher publish more, publish better and publish faster.
This is all about, you know, the opportunity about the reduced cost, increased revenue and also the share the more meaningful, useful information in the industry in the world. So this is already, it's a big opportunity. So Yeah. Great Well we're coming to the end here. So one maybe last question. Any books, websites or leaders in the area that you'd recommend our attendees sort of read follow to find out more information.
Any any sources that you love. So do you do you have a suggestion or I'll start with just say SSP Webinars are a great place. Yeah no, I don't know. I mean, it's I think a lot of just experiencing this and trying out tools is a great way. And then there's a lot of people. You know, making these big proclamations of what this is and what this isn't going to be.
I think going to the literature is great, right? We have a lot of answers. We are doing a webinar where it's like librarians looking at how chatgpt insight was used in the classroom. And it's interesting. Right? and so there's tons of different places you are yourself maybe publishing some of them on this call, which is fascinating.
And I think those are important because there's actual analysis and evidence behind them, like everyone's going to say we got the fastest, best tool. And we look at this benchmark. But how does that impact editorial? How does that impact publishing. And undoubtedly there's more and more studies starting to come out on that. So look within.
Yeah that's great. Anything else you'd like to add? Any hung or so. Yeah for me I usually the go to I found the Scholarly Kitchen is a good place to this. Although maybe because I'm also the chef of the one chef of the scholar Kitchen. So I introduced this and but this is really, you know, the thoughtful, you know, the things about, you know, the scholarly publishing there.
And also, as Josh mentioned, keep reading the research papers about the AI, the peer review. There's many things. And also, I usually go to the media that the, you know, the it's called the media. So it's very good, you know, the place to understand the many, you know, the deep complicated the, the concepts. And also I use AI tools like chatgpt perplexity to help me summarize, you know, what happened in the world.
Yeah great. Rocio, anything to add or. I think you lost your audio again. Sorry can you hear me now? Yes Yeah. Yeah great. I was just going to reiterate resources like SSP, the Scholarly Kitchen, but also just the CAC meetings.
Like any like big industry organizations are putting out really good content and that helps keep up with. Yeah, the changing landscape. And honestly, like I think Hong said it, but ask chatgpt to give you a summary of what's been in the news recently, and it's really good at doing that. That's great. Well, Thank you all.
This has been a really engaging discussion. I really appreciate your time today. And to our attendees, Thank you for your questions and for being here this morning. So you will receive an evaluation form by email within 24 hours. And a link to the evaluation form has been will be posted in the chat. And we encourage you to provide feedback to help us shape future programming.
And you can also visit the SSP website for information on future webinars. As you know, the discussion was recorded. All registrants will receive a link when posted on SSP website. You must be logged in with your SSP login credentials to view. So thank you all for joining us today and this concludes our session. Thank you.
Thank you. It's great.
Form Title Download
Transcript
Form Title Bookmark
Entire Media
Segment(s)
Custom Clip
Start At:
End At:
Form Title Share
Entire Media
Segment(s)
Custom Clip
Start At:
End At:
Form Title Accessibility and Keyboard Shortcuts
The cadmore media player is a professional video and audio player experience designed to deliver media as well as relevant transcript, segmentation, and metadata. On the screen is a traditional html video player with buttons such as play, pause, forward, captioning, language and play speed selection and more. Adjacent to the video player are generally tabs, one of which shows a scrolling transcript of the video and the other shows any segmentation of the media, sometimes referred to as chapters. At the top of the video is a menu bar containing an Explore button and Tools button. The Explore button allows you to dig deeper into the video by opening up an explore dialog which contains tabs for a visual navigation of the video using thumbnails, an about tab which contains rich metadata about the media, a segments tab which contains further information about the video, and a related tab which contains links to other media related to this media. The tools button opens a menu with options for sharing the media on social media, creating bookmarking links, creating embed codes for your website, generating citations, and more.
Keyboard Shortcuts
Keyboard shortcuts only work while you are in forms mode. By entering forms mode, you will be able to quickly access portions of the media player with single key commands.
Spacebar: Play/Pause Toggle
I: Info Menu,
O: Tools Menu
T: Transcript Tab,
S: Segments Tab
F: Forward 15,
R: Rewind 15
L: Languages,
P: Playspeed
M: Mute Toggle,
V: Volume Bar
N: Video Only Toggle,
B: Full Screen Toggle
C: Captions Toggle
Escape: Closes dialogs and menus / Exits Player
Z: Next Transcript Chapter
X: Previous Transcript Chapter
Q: Next Transcript Paragraph
W: Previous Transcript Paragraph
Entire Media
Segment(s)
Custom Clip
Start At:
End At:
Embed Size
-
You Size - Responsive
-
We Size - Responsive
-
Picture Overlay Popup
-
400 x 225
-
512 x 288
-
400 x 700
-
768 x 1024
-
1024 x 512
-
1280 x 608
Form Title Cite
No citation provided.
Info
Segments
Related
Thumbnails
Exit Clip Mode
You have requested to see a portion of the video outside of the clip.
Click OK to switch to the full video or click if you would like to stay inside the clip. You can always go back to the clip with Tools and Clip Mode