Cadmore media player playing video Understanding AI Traffic - Bots, Crawlers, and What They Mean for Your Platform
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:
Understanding AI Traffic - Bots, Crawlers, and What They Mean for Your Platform
Description:
Understanding AI Traffic - Bots, Crawlers, and What They Mean for Your Platform
Thumbnail URL:
https://cadmoremediastorage.blob.core.windows.net/6577f711-61fa-4109-a56f-502045d93439/thumbnails/6577f711-61fa-4109-a56f-502045d93439.jpg
Duration:
T00H55M16S
Embed URL:
https://stream.cadmore.media/player/6577f711-61fa-4109-a56f-502045d93439
Content URL:
https://cadmoreoriginalmedia.blob.core.windows.net/6577f711-61fa-4109-a56f-502045d93439/PSWS26 April Recording.mov?sv=2019-02-02&sr=c&sig=c58xfBqPh89W9OVjqYHdZLwyrn%2FE4mnxyMbNUJwokGI%3D&st=2026-03-26T21%3A26%3A58Z&se=2026-03-26T23%3A31%3A58Z&sp=r
Upload Date:
2026-03-26T21:31:58.5849916Z
Transcript:
Language: EN.
Segment:0 .
STEPHANI LOVEGROVE
HANSEN: Welcome.
HANSEN: My name is Stephanie Lovegrove Hansen. I'm the VP of Marketing at Silverchair. And thank you for joining us for today's event. This is the first in the 2026 Platform Strategies webinar series. We are really looking forward to today's discussion. But before we get started, I'm going to cover just a few logistics. So this year's webinar series looks at the ways that the definition of the end user has changed in recent years and how we can design platforms and strategies that effectively serve both the human and machine users.
HANSEN: So this free webinar series features three events, and you may register for the next two events on the same page as you did for this one. OK, there we go. Those are going to be on April 14th, we'll cover on platform discovery. And May 12th, we'll look at MCPs and agents. We designed these webinars to feel more like a discussion. So we encourage you to get involved via the chat, via the Q&A feature.
HANSEN: And we will have some time at the very end for Q&A. The event is being recorded, as the voice told you at the beginning. And a copy of the recording will be made freely available to you on our website and via email following the event. Finally, at the end of the event, you'll see a short survey requesting your feedback. And we appreciate you taking a moment to just let us know what else you'd like to see in future events.
HANSEN: With that, I'm going to hand it over to the moderator of today's event, Hannah Heckner Swain. HANNAH HECKNER
SWAIN: Hi, everyone.
SWAIN: Thanks so much, Stephanie, for that intro. As Stephanie mentioned, I'm Hannah Heckner Swain. I'm the VP of Strategic Partnerships at Silverchair. And I'm really happy to have these esteemed panelists joining me today. Before I introduce them, we are going to open up with our first poll. There will be one at the end as well. So if we could just launch that poll. And it is going to ask you how your organization has experimented with blocking or allowing certain AI bots.
SWAIN: So you could indicate yes, no, you're planning to look into this, or you're perhaps not sure about what your organization is doing, that would be great. As folks are doing that, I will introduce Lou, Paul, and Robb. We have Paul Gee with us, the VP of Product at the JAMA Network, Lou Peck, CEO of the International Bunch, And Robb Burgess from Silverchair, our VP of Technical and Security Operations.
SWAIN: So a great group here today. How are our results coming in, Steph? Are we ready to see? OK, great. Got some hearty yeses and then a mixed bag. So this will be great. We'll ask a question at the beginning-- at the end of the session to see how your thoughts may shift moving forward.
SWAIN: So we figured it would be really helpful to ground today's discussion with a look at some statistics. So what you're seeing here is usage by week that we, as Silverchair, the platform, are seeing on our outer layers of infrastructure. I'll pass the mic over to Robb to speak more about what this is telling us. We thought it would be a helpful place to start today.
ROBB BURGESS: Thanks, Hannah. Yes, so this is a combined view of all of the Silverchair platform traffic for the first 10 weeks of 2026, or yes, 2026, I guess. The left-hand, light blue, is what we see as human traffic to the platform. The middle column there, the kind of darker blue, that is traffic that we have served to known bots.
ROBB BURGESS: So that could be Google bot, Bing bot, some of the good crawlers that we let in. Within these percentages is also some of the clients have spiked on testing out letting in the AI bot. So that would be accounted for within this percentage as well. And on the right-hand side is the amount of traffic that we have not served. So there is an extremely small amount of traffic that Silverchair will outright block.
ROBB BURGESS: And that's going to be malicious traffic. So someone trying to perform a SQL injection or cross-site scripting. And we're just going to absolutely block that you can't come in. The vast majority of this traffic that we're showing is not served is some sort of automated system, say a bot that has not been approved or a crawler that has not been approved or some other automated system that attempted to go to a site but then could not pass the managed challenge.
ROBB BURGESS: I'm sure everyone's seen that Cloudflare pop-up. You go to a website, you get the little spinny thing, Cloudflare says, checking that you're a human, and then it goes away, and the site pops up for you. That's the vast majority of the traffic that we're not serving is just automated systems that aren't going to pass that kind of human check. So, yeah. And getting into the bot types, as I mentioned, most of the bots in that middle column that we are serving are search bots.
ROBB BURGESS: Your traditional Google bot, Bing bot, Apple bot, things of that nature that's indexing sites to power search around the globe. Some of the newer search bots that we're seeing are for these AI platforms, but they act exactly the same that Google bot does. So you have, say, OpenAI, they have their own AI search bot. And it's doing the exact same thing that Google bots doing.
ROBB BURGESS: It's indexing the publicly available content or web pages on sites, so that when someone using OpenAI or ChatGPT, it's just got a list of-- it's got an index. So it's not having to go search on its own. When you search for something in ChatGPT, it's got its own kind of search index there for you, just like Google does. There are the training bots or the AI models that go train their large language models.
ROBB BURGESS: We block those outright for all client sites and all traffic. We recommend that you keep doing that. You don't want these LLMs training on your content, for the biggest reason of that you lose control of the content. Like, I know it might be rare, but if you release a correction to some article, you don't know what the LLM has slurped up. If it slurped up the previous one, that is incorrect, is it going to pick up the new version that Is correct?
ROBB BURGESS: Maybe, maybe not. But we just recommend that you keep blocking these. And then the user bots. This has been where some of our clients have been experimenting. So this is where, say, I'm using ChatGPT and I search for issues around knee joints. And it goes to one of our client sites and pulls information back from it. That's just going to give me information about that article, say.
ROBB BURGESS: And then it will also give me a link to say I got this content from xyzpublisher.com/kneejoints, whatever it may be. And so that's where we're seeing a lot of the referral traffic come in. So traditionally, Google Scholar has been at the top of the referral charts. And we're seeing the ChatGPT user referral creeping up to be comparable.
ROBB BURGESS: In some cases, it's now the sixth, fifth, fourth highest referral we're seeing. And so this is, as we say, the strategic decision. How do you want your content to be disseminated? Do you want to experiment with AI bots? This would be the place where you would do it with these user bots.
HANNAH HECKNER SWAIN: Awesome. Thanks so much, Robb. So I think it could be helpful to start our discussion today with the publisher in the room, Paul. So as the JAMA Network looks into these sorts of experiments and you all look at the usage from humans and not humans coming to your platform, what are your big focuses in these meetings and what have surprises been looking at your non-human and human traffic?
PAUL GEE: Well, the surprises, I guess, have always been the same. Like we always-- I mean, for 15 years, we're always having spikes around different articles. And sometimes they're human, and sometimes they're not. Sometimes they're from geolocations that cause suspicion and they're human. And sometimes they're bots. Taking a macro view, I feel those investigations are going to continue and always happen.
PAUL GEE: When we're talking about the bots, the thing that is always a balance is we want to be able to serve the content, not just for the business model, but for the mission. JAMA should be something that people have access to is a general idea. And that chart that shows a 53% to 55% turnout rate, really is alarming to a publisher who's trying their hardest to disseminate.
PAUL GEE: That's our job on the publishing side when we're not in editorial. So I think that was the most surprising thing that I've heard and that we've discussed. And when I brought that back to the publishing group that we're turning away 50% of the traffic, everyone was like, well, then there's definitely a margin of error in there that we can benefit from looking closer at.
HANNAH HECKNER SWAIN: Got it. So do you think that, when you decide to selectively allow certain AI traffic in, what were the main tests? Or was it OK, we're blocking these 53, let's be more permissive and then winnow down? Like how did you all make those initial decisions?
PAUL GEE: We followed the advice of people at Silverchair and also people at an SEO vendor. And we looked up and read the definitions of those bots, those user agents, and thought about what they really do. They have a search agent and a user agent. The search agent performs for GPTs, the same thing that Google bot does. It's keeping a wide cache of everything that's available on a daily basis.
PAUL GEE: And then the user agent can be served quicker, the right content. That's, to me, no different than a Google link. And so we kind of took that approach that if our job is dissemination and discovery and allowing aggregators that honor our business models to play that we don't really-- that we actually have a pretty good case to start testing, letting them in. And when we did that, we looked for signs with Silverchair and in our own analytics of any different levels of exploitation than are there already and didn't really see them.
PAUL GEE: We know that we benefited a great deal from Cloudflare when we installed it, because it did eliminate a lot of bot traffic. And when we released this again six months later and started letting partitions in, we didn't see that same return. And we do see healthy levels of page views per session from the sources that are GPT, that are the equivalent of Google and other preferred sources.
PAUL GEE: HANNAH HECKNER
SWAIN: That's great.
SWAIN: Now, Lou, you work directly with publishers. And as a consultant, what experience have you had with guiding publishers exploring this process about where they find the right spot between being too restrictive or too permissive?
LOU PECK: Yeah. It's a real worrying concern, and it depends on the type of publisher. So if they're a society publisher, not-for-profit, commercial, how small they are, how big they are, university, press, and what resources and knowledge that there is, we tend to find that there's a real knowledge to action gap. So we know that AI is here. We know that it's been around since the dawn of computing. We're just a lot more aware about it now.
LOU PECK: And we're now just fully grasping the capabilities that AI has. And so we're finding that some publishers are really paralyzed by scraper's dilemma. And what does it really mean? And what content should we have, and how is that discoverable? And what we let in and what shouldn't? I'm going to talk about some of the things later, but certainly in terms of trust and credibility, that's incredibly important as we move forward in what we do and how different search engines programs and services view how we appear online.
LOU PECK: Google alone last-- I think it was last week or the week before, launched six new tools. And in terms of the user experience and the things that those different tools can do now, it's quite scary. And Google Scholar has a lot that's going to be coming, too, in terms of this is it's year turning from traditional search to more semantic way of doing things.
LOU PECK: So it's very much dependent on the organization, what their core goals are. But really, publishers do an incredible job of ensuring that there's credible research out there. There are workflows and processes really to ensure that what we're publishing is trusted. And there's a lot that goes on with research integrity. But then after that, you've got to make sure that people read it and they discover it.
HANNAH HECKNER SWAIN: Yeah. Those really important core goals of discoverability, reputability, maintaining value. Are there any guiding principles or success stories that you have from working with publishers, as they consider this, as they perhaps open new value streams with these decisions? I know we're still early days with commercializing this kind of traffic, but any sort of experiments or guardrails that you have established in your working with publishers or thoughts?
LOU PECK: Yeah. There's certainly a lot in terms of brand dilution and content devaluation. So there's concern that bots are going to ingest high-value content, and then they're going to mix it with some of their hallucinations, AI hallucinations. And so it's going to make it lower quality. And so in terms of what we're showing readers in terms of what's truth and what's noise, and that's definitely going to be harder.
LOU PECK: What we're seeing that there's a lot of diversification in our industry and how societies or how publishers can generate other types of revenue. And so we're seeing a real movement in terms of advertising on platforms as well and embedding third party partners who can help them to do that. So in some respects, there's going to be different types of traffic that people are going to be wanting to come in.
LOU PECK: So we focus too much sometimes on quantity versus quality. And actually, I think I would rather use an AI tool to do a thousand searches-- to do searches for a thousand scientists, rather than sending a thousand people who are Joe Bloggs, some layperson, to my page to look at stuff, because that really inflates the statistics and what we're looking at.
HANNAH HECKNER SWAIN: Yeah. Thinking about that quality metric, it can be really hard to dig into that. You're not always sitting next to someone who's looking for and finding your content and reading it. So maybe I'll start with you, Paul. And Lou, feel free to jump in as well. Like, how have you dug into the user experience success of looking at this data that's coming in and how can you discern a good experience is happening?
HANNAH HECKNER SWAIN: You mentioned the click-throughs and time on site that you're seeing from the user agents, but is there more deeper dives that you're looking at or additional data points?
PAUL GEE: Honestly, not right now. Other than, does it fit the mix of real human traffic? And if so, should we let it in? Where we're focusing a lot of our attention is building out tools and services for those that do want to use applications. So we're pretty progressive, while we're also thinking very heavily about security. And we keep those in a balance.
PAUL GEE: But I feel like that's the same-- that's the practice that we've had with Google or with other search agents for years, that healthy dilemma of do you work with Google Scholar like 10, 15 years ago, and how can you not? And you end up in a situation where you're collaborators with these companies that are enhancing discovery. So I see institutional traffic coming through these channels.
PAUL GEE: I see high usage and high propensity to stay on the site. So with those three things, it's the fingerprint of what you want to see. And the more qualified to lose point, the more qualified-- if what these tools do is qualify your traffic for you so that it's set up to deliver an experience that's better for the user. And then they're more engaged, they're more likely to site and the traffic's high, then we like-- it feels like a win-win.
PAUL GEE:
HANNAH HECKNER SWAIN: Now we've talked a lot about value, communication, keeping your brand strong, making your content discoverable as decision points in this journey. Robb, do you want to maybe speak about the backend stuff that publishers need to be considering when stuff like this comes up, with regards to infrastructure and things like that?
ROBB BURGESS: So there is a discussion to be had when a publisher initially wants to start experimenting with allowing these AI bots in. For some publishers, it's been a slight increase in traffic, 5% to 10% maybe, which is entirely absorbed within the current infrastructure that Silverchair holds for any given client site. For some publishers, we have seen between 25% and 50% traffic increase just from these AI bots trying to scrape-- not scrape the site, but access content on the site.
ROBB BURGESS: And when it is that level of a traffic jump, we do have to have discussions about protecting the current infrastructure, whether we want to work with a publisher, to talk about a project to expand it. We've put rate limiters on this type of traffic for some publishers that have had a large traffic increase from it. So that's kind of on the infrastructure side.
ROBB BURGESS: And then looking forward, we're also exploring the possibilities of serving this AI traffic in a manner that a machine likes to read. Right now, if you go to any of the publisher sites, go to an article. It is designed in such a way that it is readable and digestible for most of us. And it's not that readable. It is readable and digestible for machines, but it is not optimized for machines.
ROBB BURGESS: And so just removing all of the CSS and all of how it's designed, and just having the content in somewhat of an XML markup format, serving these AI bots just that information is something we're exploring. It's going to lessen the load on our physical server infrastructure and give this information in a manner to these machines that like to read it that way.
HANNAH HECKNER SWAIN: Great. Super helpful. There's some questions that have come through from the chat. One of them is touching on one of the slides that we showed at the beginning in regards to allowing search bots in. There was a note that says block these and you disappear from search results. Should that have perhaps said block these and you disappear from chatbot responses?
HANNAH HECKNER SWAIN: I think the answer to this is no. It's really just allowing the search bots in keeps your content indexed so that it's easier to pick from. Is that a correct statement, Robb, first?
ROBB BURGESS: It's somewhat nuanced. That statement is true. If you blocked a Google bot, you're going to be removed from the Google index. If you search for your site on google.com, it's not going to come up. Now it's somewhat of a different scenario with these AI search bots. For example, ChatGPT or OpenAI, if you ask it a question, and it has to search around the internet for something, it's going to use Microsoft Bing search just like anyone can type up bing.com and search for something.
ROBB BURGESS: And it's going to be a bit slower. It's going to burn more of your tokens if you're a paid user of ChatGPT. But if you have allowed that OAI search bot to index your site, instead of using Bing to search the internet just as is, it's going to use that searchable index. So it's going to be faster and in theory, give the user better results. HANNAH HECKNER
SWAIN: Thanks, Robb.
SWAIN: A recent question that just came through, which I think is hewing back to just the vocabulary that is part of this overall conversation, and it's about the difference between a bot and a crawler. A crawler is a type of bot.
ROBB BURGESS: Correct. So, yes. Not all bots are crawlers. All crawlers are bots.
HANNAH HECKNER SWAIN: A square-rectangle situation.
ROBB BURGESS: Generally, we refer to a crawler if it is something that is trying to consume all of the content. It's programmatically going through all of the pages step by step to consume everything.
HANNAH HECKNER SWAIN: Thank you. Another question that's come in. Not all AI tools provide a link with their responses to users. What kind of ongoing research, the questioner asks, is Silverchair doing-- I think this is also a publisher-- question regarding evaluating the referral value of each bot? What guardrails exist for bots that do not provide robust click-through rates? So I think maybe breaking this up into questions.
HANNAH HECKNER SWAIN: The AI citation counts. Lou, Paul, how have your organizations or the organizations you're working with, how are you all looking into those AI citations? And how does the use of citations in answers-- how does that come into your decision-making when choosing which bots to use?
PAUL GEE: So I don't think we know enough yet. If we're talking about traffic, Right now, we're only allowing in a subset of the GPTs that have a good reputation, and we're allowing the search and the user agents to index and then therefore serve our results. And when we look at that traffic, it looks healthy, like I was I've been saying.
PAUL GEE: How it's used and whether it impacts if we're talking about bibliometrics and citations, we won't know for a couple of years. I think one strategy here is to ensure that we don't take a dip in citations because we're aggressively blocking. And it's a balance of trust. I will say in the beginning, like in 2024, we were probably the first publisher to do a full hard stop robots block for every known AI crawler.
PAUL GEE: We did a wild card. And we did not release it until there was information that we could control it and identify which bots we were letting in, and then we only permit it in three or four. A few of those drive, and one of those really, ChatGPT, drives the lion's share of the traffic. And I believe, I mean, I'm guessing that certain providers like Claude, I don't believe it's because we're not there.
PAUL GEE: I believe it's because they probably have a business model where they want publishers to license their content directly through different channels, which is a growing model that that's coming up where they are starting to pay to play. And we'd be looking at that.
ROBB BURGESS: Yes, I can confirm with all of the publishers that have experimented with letting in these AI bots, 99% of it is ChatGPT. Less than 1% is Perplexity, and less than 1% is Claude.
HANNAH HECKNER SWAIN: Yeah. ChatGPT just sort of became I'll google this, I'll ChatGPT this, that market share thing. Lou, you were going off of mute.
LOU PECK: Yeah. Doesn't roll off the tongue as easy as I'll google that. Doesn't mean that I'll ChatGPT it. I mean, I get confused about acronym all the time. Yeah, I was just thinking. So in terms of how we're measuring things, that's really important. We have traditionally measured things in certain ways and we need to rethink that. So when we think about a bot comes in and it crawls a site and it's doing what it needs to do, it takes like half a second whatever it's doing.
LOU PECK: But because there's no continue click, it's not being registered. So it's registered as a zero click. So in terms of how we're measuring things, and so rather than looking at things like keyword rank, we could be looking at citation share as well. There are different metrics that maybe we as a community need to look at. And I think that NISO and-- who is it?
LOU PECK: Let me just quickly check. I think that there's something going on between NISO and COUNTER, where they're looking at a metric called Access_Method agent. And so really being able to differentiate between human usage and then what's coming in otherwise is going to be incredibly important. What I worry about is I've been doing some research this week speaking to libraries and is the problems that they're facing, which they face every year, which is like a 10% decrease in collections budget or overarching budget for their library, and how that's going to affect publishers and service providers moving forward with subscriptions or different open access models.
LOU PECK: They're pulling apart transitional and transformative agreements. So I'm concerned about the value that they see in the content that they're given in the way that it's currently being given and what that actually means. And part of the way that they measure, there are two primary ways of measuring is cost, but also usage. And so how they measure that usage, is that still relevant?
LOU PECK: Do we need to have a complete change in how things are measured and reported? And that's what I worry about for the future with publishing.
HANNAH HECKNER SWAIN: Yes. Those new counter standards can't come quick--
LOU PECK: Yeah, absolutely. And just in terms of what's been explored and who's using things. And then when people come to your site, you don't want Joe Bloggs just doing a search and I don't know, I've got a mole on my back, and it ends up pulling up some credible research data. But you want the researcher who's actually looking for their research to then come into your site and then experience the content. And that's where we talk about quality versus quantity.
LOU PECK: It is about the quality. But that would mean less quantity coming in. So tricky, eh?
PAUL GEE: I would say I think that we're going to see more and more applications and businesses that are within our space. I think that's only begun. I mean, you can see we've been working with OpenEvidence. That's an example of a very precise market opportunity that's in our area. But I only see more of those growing or being launched. LLM technologies and these technologies are becoming more easy to launch, and they're going to grow up in all of our spaces.
PAUL GEE: Probably medical, it will happen first, in a lot of ways. But it's going to hit all areas of STM. And I think that will be a growing business line for most groups as you begin to license content in new ways, but also you get significant traffic back when they create link back services correctly. One thing that I've noticed, not all GPTs present the links backs the same.
PAUL GEE: Some do it very, very well, like OE and GPT do it pretty well, but others they create layers, where you have to click. And then you get a slide out that then tells you about the thing you clicked on. And then somewhere in there, there's a link, like they're really making it hard to leave that environment, which I tend to think will get them dinged with users. Over time, the users always win.
PAUL GEE: If you want an article, you're just going to either stop using the interface and go somewhere else, or they're going to change the interface to get you to the article.
HANNAH HECKNER SWAIN: Great. OK. We're getting some questions in. Trying to think of the best one to discuss next. OK, I'm going to combine two questions. Just a two-parter directed initially at YOU Robb. When we look at referrals, is Gemini traffic separable from Google Search traffic? So that's the first part. I imagine that will hopefully be a short question.
HANNAH HECKNER SWAIN: Are we able to tell the difference?
PAUL GEE: We are not. Generally, Gemini is going to use all Google bot indexed information.
HANNAH HECKNER SWAIN: Got it. And then in thinking about the use of agentic browsers, how do these fit into the conversation?
ROBB BURGESS: This is a bit of a curveball, and we are not sure what that's going to look like. So I have been experimenting with the claude version of-- it installs right on top of Chrome. And I can go to any website that I have access to. Log in and ask Claude to summarize the data, do whatever it wants to with that data. And from a infrastructure web server point of view, there's no difference between me as a person using Chrome and going to xyz.com, and me using Claude within Chrome to go to xyz.com. There's no difference as far as the server logs can tell you.
ROBB BURGESS: The only way you could do that is if you told Claude to say, scrape a site, and it's clicking through different URLs on a page quicker than it is possible for a human. But other than that type of heuristic analytics, you can't tell the difference.
HANNAH HECKNER SWAIN: And while we have you, another sort of vocabulary question. If a piece of software focuses on just a single site to extract all of its data, is that still considered a crawler? Or should it be classified as a site-specific bot?
ROBB BURGESS: I would consider it a crawler. And if it's approved, say, Bing bot, Google bot, great. I mean, we want to be good partners with all of our clients. And if you tell us you want X, Y, Z crawler to access your site and grab all the information, we're going to do our best to facilitate that for you.
HANNAH HECKNER SWAIN: Great. Thanks, Robb. Paul, this is initially directed towards you from Allison Bellin at Duke. So thinking about specialized AI and chatbot applications, they're proliferating, built to support scholarly research and specific disciplines. So when we think about these bots accessing your content, do they manifest in-the-traffic data? Or are these-- as drama experiences, are these separate deals where it's licensing that you then get usage from that?
HANNAH HECKNER SWAIN: What does this usage outlook look like as JAMA explores these? The way that I'm talking about it here is usage of the site due to referrals back from these discovery portals, I'm thinking, especially in the context of how I've been answering the questions, I'm thinking about them as discovery tools that bring traffic back. And we are interested in getting usage statistics and figuring out how to handle that from off-site aggregators.
HANNAH HECKNER SWAIN: And I think part of that is with a group like ResearchGate, you can get counter integrations back and there's a model for that. We're really waiting to lose point for counter to catch up with this type of solutioning. And then I tend to think we'll have-- we're starting to get all the tools we need in order to do our traditional dissemination work just with this new type of technology, which I think, at the end of the day, Robb, when you saying agentic browsers, regular browsers look the same or feel the same, I wonder how much like this becomes noise in the end.
HANNAH HECKNER SWAIN: We're simply just delivering through technology the content that we create.
HANNAH HECKNER SWAIN: Yeah.
ROBB BURGESS: I think that's absolutely right. You're going to be delivering the content in the manner that you've specified, whether that's the general public going to ChatGPT and trying to access your content that's in front of the paywall, or you have a researcher at some institution that has a license for your content and are using the browser, the agentic plugins. It's basically the same thing. You're disseminating the content in the manner that you've specified to the public, whether that's in front or behind the paywall.
HANNAH HECKNER SWAIN: As we look at that disintermediated future and organizations like the AMA wanting to serve their users' needs where they have them, how do conversations about the JAMA brand-- what does that look like in that future? There's obviously editorial rigor, but if you're not holding up a print magazine, how do you communicate that brand?
PAUL GEE: That's what we're all learning. I mean, not extending your brand, I think, is a problem. And it would be a problem for anyone. If your brand isn't known in the future through some of these tools, where will it be known? Print was the way. In many ways, people have said, we don't read print for years. But print was this pamphlet almost or this reminder that the journal exists, so that when you saw the result in Google, you recognized it and you clicked.
PAUL GEE: And that glue, that intangible glue, it comes up in customer research, it comes up in all these ways. It's not deterministic. You can't analyze it in the data, but it's there. We're humans and we live by recognizing brands. So I think we need to continue to have rigor around the customer research we do and the evaluations we do of how people recognize us over time.
PAUL GEE: I think we all need to as publishers. And make adjustments as we go. We need to be mutable and change day to day, which is how, I think, this game gets played.
HANNAH HECKNER SWAIN: Great. Great points. Lou, I'm going to start with you on this question. It's from Louise Russell. Do you think there will be a shift in funding from library budgets to the institution's technology budget for AI licensing deals? Just thinking this could have implications for how we and our customers measure value.
LOU PECK: Well, sadly, as a librarian, I can't talk so much to this. But certainly, that's a great question. Thank you, Louise. Yeah, it's interesting. I think that-- well, I think the losing budget anyway. So shifting money that isn't there anymore is going to be practically impossible. Shifting money into innovative ways that they're looking at how they can license content, and maybe they might do things like per-article views, things like that, actually only just going for the articles that people are really interested in, which is going to completely causes problems when we look at usage data across the board.
LOU PECK: And because people are only going to be having access to things that they need rather than us bundling things together. So libraries are definitely doing a lot in terms of their use of AI, how they're using it for their user experience, how they're using it in their technology. We've seen that there are universities that have been created in the Middle East, which are very AI-specific.
LOU PECK: That's what they've been created for. So there is a huge amount of effort and there is change happening. But for libraries, it's about, how can they get the content that they need in the best way possible, for the best price that they perceive as value? And that value point is really important. And how they value our content is quite subjective. And a lot of them measure our content in very specific ways, very traditional ways that they have been doing it for years and will continue to do so.
LOU PECK: So that's potentially problematic. But to say to Louise's question, yes, potentially. But I think there's lots of things that they're looking to do with their money that has been coming historically to us in this industry. HANNAH HECKNER
SWAIN: Another bit
SWAIN: of a crystal ball question from Stuart Maxwell. Could there be future cases for researchers to evidence that they have actually accessed underlying content, rather than just relying on agentic outputs? So you can't cite Wikipedia in your paper argument here. What might that look like in the future?
PAUL GEE: That's what DOIs do.
HANNAH HECKNER SWAIN: Yeah.
PAUL GEE: I don't know if there's-- I like the question because it made me think like about if there's a way to fingerprint these and use that. And I mean, if there's a technology that allowed citations to be validated whenever or wherever they went, then that would be used, I'm sure, by all the researchers who want to be able to trust their sources. I think trust goes two ways.
PAUL GEE: I can't imagine that researchers feel a lot of trust. They're excited that they can get information quickly. But how do you validate what you're seeing with these new discovery tools? But it's also compounded. I mean, the amount of content that's published in the last 25 years, year on year, the increase is exponential. At a certain point, we simply need to make discovery easier.
PAUL GEE: And these tools do that. So finding ways to pinpoint authority and marks of integrity and trust are equally important.
HANNAH HECKNER SWAIN: Yeah. And this brings to mind a conversation that we had in one of our prep calls about how growth in AI usage could actually be a huge democratizing factor for access and research. Lou, you had some good points on that. Mind sharing a bit with the group?
LOU PECK: Yeah. No problem. So one of the things that I just want to mention is that AI is the way it looks at brand, because we were talking about brand, is that it sees brand as a fact, whereas we see it more as a value. Guys, sorry. Sorry, my kids have just come in honestly. And so we think about it as a logo, an experience, what's it look like, what's it feel like.
LOU PECK: And it just doesn't look at us like that. And so it looks to a brand as an authoritative answer. And it's about just what Paul was saying, what we said before about trust. And so when we think about what's coming into our site, what's linking to our sites, what's linking to our content, all those spam, all those naughty sites that are going, trying to be credible themselves, building up their trust score, their Google trust score.
LOU PECK: You can actually use a Google tool to reject them, to actually cut them off. So a lot of the time, we're building things with our eyes. And so we really need to be building things for intelligence. We optimizing for a bot is actually optimizing for clarity and accessibility. And we know how incredibly important that is. So if a machine can't understand our research hierarchy or our metadata, then actually, how can someone like myself, who's neurodiverse or anybody, really, how can they using assistive technology if they've got a disability or differing ability?
LOU PECK: They probably can't either. So we need to really be thinking about the future of what we're doing, how we're presenting data. Is it in other ways rather than just in a PDF or HTML? Is it in a video? Is it in audio? How are we disseminating that important information? It's much easier to present a video of an experiment than to write about it.
LOU PECK: So it isn't just about links anymore. It's about how a map of research relates. And so our platforms need to be bilingual. So you've got this beautiful layer, the Silverchair, and underneath all the other stuff that Silverchair is doing. But it's about what's engaging for a human and then also the AI that they're using. So it needs to be clean, structured, machine-legible, but also, for us, ourselves, about how accessible is that content.
LOU PECK: And we need to be thinking about not being where we want users to be, but we need to be where users are, because we can't force people to do the things that we expect them to do.
HANNAH HECKNER SWAIN: Yeah, definitely. Great. Let's see. We'll have two fun questions to close this out, and I'll open the poll after this first question. So as evidenced by this conversation, even though people that think about this all the time, we're doing lots of experiments, we're learning a lot, but we're still doing things in real time. So with this fast-moving environment, what do members of this group wish they had known a year ago that-- like things now that you wish you knew a year ago, and what are you trying to figure out?
HANNAH HECKNER SWAIN: Robb, I'll start with you.
ROBB BURGESS: Seems like a loaded question. No. I wish I had dug deeper into the underlying infrastructure of the large language model, the RAG retrieval models and how that all works. I think I would have been a bit further along in the journey than I currently am.
HANNAH HECKNER SWAIN: Anything we're still trying to figure out? I think a lot. What's your number one thing?
ROBB BURGESS: Just how to use it effectively. But there is always more learn. And using it from both sides, maybe I don't know if offensively and defensively is the correct term, but on one sense, from part of my role is to protect the Silverchair platform. So being defensive on that end from unapproved AI bots, crawlers, all those good things, but then also being on the side of what we need to employ it to do better, to do things faster, to make sure that we are moving along with the times, that we're keeping up to the cutting edge within our platform and for all of our clients.
ROBB BURGESS: HANNAH HECKNER
SWAIN: Thanks, Robb.
SWAIN: OK, Paul, what do you wish you had known a year ago?
PAUL GEE: Well, what I wish I knew better now, and should have started a year ago, is just understanding how to ensure that as we allow this traffic in that we're getting it to the right content. And I was thinking when Lou was talking about-- something you said made me think about the value of SEO. Like when you initially started doing SEO, everyone would say, well, you're writing that copy for the bots, like a blurb at the top of a page that explains what the page is for.
PAUL GEE: It's actually for humans. It makes your sites better. If someone can actually understand the context before they dive down to the bottom. And I see us starting to go down the same routes. If we end up making special types of metadata around our articles or interfaces that people don't see, but the bots consume, it stresses me out a little bit that we're making better content for the bots than we are for the humans.
PAUL GEE: As publishers of primary literature, we prefer the primary literature, but humans really do enjoy context. And I wish you learn these things real time, but as I see that coming up, I think that's a really fun territory where Publishers content could get better not just under the hood as they try to feed the bots better sources, but using it within the interfaces to make the experience better as well.
HANNAH HECKNER SWAIN: Yeah. Where the search engines optimizing content, that was written best for human consumption. And are the bots doing the same? So we should just human optimization, it becomes all one thing. That's great. Lou, what do you wish you had known a year ago?
PAUL GEE: That clicks would become a vanity metric. So we've always been thinking-- well, not always. We have recently been thinking of AI as more of a discovery tool. So that leads people to sites and they can view the full paper. But actually, it's more about they don't want the paper, they want the insight, which goes to what Paul was saying. And so search referrals, we've seen them plummet by about 33% to 55% in the last year.
PAUL GEE: So I think it's really better understanding that metrics are going to change. And actually, it's not about optimizing for traffic. That's what we've traditionally done traffic, but start optimizing for inference interference present. So ensuring that our content is the preferred source for RAG, that it's trusted, that it's credible, and that we continue to really support with everything that we do already for that content, that we continue to support that in a way that it remains discoverable with so much noise, because there is an incredible amount of noise out there.
HANNAH HECKNER SWAIN: Great. Thanks. OK, Steph, I think we're ready for this final poll. And as folks are filling this out, so how has this session changed your thinking on bot access to content? Indicate your thoughts here. While everyone's filling that out, we've got a robust group of folks attending this call. So I'm going to go around and ask all of our panelists to say, one thing publishers should stop doing when thinking about AI traffic and one thing they should start.
HANNAH HECKNER SWAIN: Robb, I'll start with you.
ROBB BURGESS: When thinking about AI traffic, again, we've said this in every one of these webinars. I think you have to go back to your organization's stated goals. Is your goal to disseminate this information to anyone, everyone, all the time, everywhere, no matter what? Or are you more of the. We want to make sure that this information is as accurate as possible and serve to our authorized users.
ROBB BURGESS: There's a difference in those two statements. So I think it's going back to your mission statement and really deciding what your organizational goals are.
HANNAH HECKNER SWAIN: How about Paul, things that publishers should stop or start doing when thinking about AI traffic?
PAUL GEE: I guess. Stop having an automatic bias against bots. I see a lot of the debate about humans being preferred and bots having a lot of negative attributes. I tend to think that the humans gave the bots a lot of negative attributes, and this is it's not so easy. Humans and bots are very similar in the way that they tend.
PAUL GEE: to be communicated with and the way that we need to work through them. And I see that metaphor in a lot of different ways. But I think not trusting the bots to the point of treating them they just need to always stay outside the fence is only going to hurt the longer and longer we take that position. I think that if our job is dissemination, we need to figure out how to disseminate with these new technologies and to not fear them as much.
HANNAH HECKNER SWAIN: Great. OK, Lou, bring us home.
LOU PECK: Yeah, mine's quite similar, actually. It's about stop treating AI traffic as noise and something to be blocked. So you can't-- the future's here. The future's now. It's coming. Well, it's already here. So you're only actually going to end up making your research invisible. So start focusing on generative engine optimization.
LOU PECK: Paul talked about SEO. We love our acronyms. There's a GEO. But this is about ensuring your brand, your content, your journal, your books, whatever it may be that they are trust anchors because that's what AI models that they rely on that truth aspect. So a website is no longer a final destination. A content platform is no longer that final destination.
LOU PECK: It's more about verifying. So we need-- users need to discover the insights they'll discover in an AI interface. But when they come to you as a publisher, you need to. You need to show that it's real. So make sure that you're ready for them. HANNAH HECKNER
SWAIN: great love it.
SWAIN: Keep doing the things that have made you a respected brand. Be open to experimentation and remember your mission. Great things to end on and could be applicable to anything. So yeah, what does this poll say? OK, more open to experimenting. That's great. Wonderful. Well, Robb, Paul, Lou, thank you all so much for your insights and the conversation today. Stephanie, thank you, as always, for your wonderful organization.
SWAIN: Any housekeeping notes for us to end on?
STEPHANI LOVEGROVE HANSEN: Nope. All set. Thank you all for joining. We will send the recording around afterwards and really appreciate all your engagement throughout.
HANNAH HECKNER SWAIN: Yes, thanks everyone.
Segment:0 .
STEPHANI LOVEGROVE
HANSEN: Welcome.
HANSEN: My name is Stephanie Lovegrove Hansen. I'm the VP of Marketing at Silverchair. And thank you for joining us for today's event. This is the first in the 2026 Platform Strategies webinar series. We are really looking forward to today's discussion. But before we get started, I'm going to cover just a few logistics. So this year's webinar series looks at the ways that the definition of the end user has changed in recent years and how we can design platforms and strategies that effectively serve both the human and machine users.
HANSEN: So this free webinar series features three events, and you may register for the next two events on the same page as you did for this one. OK, there we go. Those are going to be on April 14th, we'll cover on platform discovery. And May 12th, we'll look at MCPs and agents. We designed these webinars to feel more like a discussion. So we encourage you to get involved via the chat, via the Q&A feature.
HANSEN: And we will have some time at the very end for Q&A. The event is being recorded, as the voice told you at the beginning. And a copy of the recording will be made freely available to you on our website and via email following the event. Finally, at the end of the event, you'll see a short survey requesting your feedback. And we appreciate you taking a moment to just let us know what else you'd like to see in future events.
HANSEN: With that, I'm going to hand it over to the moderator of today's event, Hannah Heckner Swain. HANNAH HECKNER
SWAIN: Hi, everyone.
SWAIN: Thanks so much, Stephanie, for that intro. As Stephanie mentioned, I'm Hannah Heckner Swain. I'm the VP of Strategic Partnerships at Silverchair. And I'm really happy to have these esteemed panelists joining me today. Before I introduce them, we are going to open up with our first poll. There will be one at the end as well. So if we could just launch that poll. And it is going to ask you how your organization has experimented with blocking or allowing certain AI bots.
SWAIN: So you could indicate yes, no, you're planning to look into this, or you're perhaps not sure about what your organization is doing, that would be great. As folks are doing that, I will introduce Lou, Paul, and Robb. We have Paul Gee with us, the VP of Product at the JAMA Network, Lou Peck, CEO of the International Bunch, And Robb Burgess from Silverchair, our VP of Technical and Security Operations.
SWAIN: So a great group here today. How are our results coming in, Steph? Are we ready to see? OK, great. Got some hearty yeses and then a mixed bag. So this will be great. We'll ask a question at the beginning-- at the end of the session to see how your thoughts may shift moving forward.
SWAIN: So we figured it would be really helpful to ground today's discussion with a look at some statistics. So what you're seeing here is usage by week that we, as Silverchair, the platform, are seeing on our outer layers of infrastructure. I'll pass the mic over to Robb to speak more about what this is telling us. We thought it would be a helpful place to start today.
ROBB BURGESS: Thanks, Hannah. Yes, so this is a combined view of all of the Silverchair platform traffic for the first 10 weeks of 2026, or yes, 2026, I guess. The left-hand, light blue, is what we see as human traffic to the platform. The middle column there, the kind of darker blue, that is traffic that we have served to known bots.
ROBB BURGESS: So that could be Google bot, Bing bot, some of the good crawlers that we let in. Within these percentages is also some of the clients have spiked on testing out letting in the AI bot. So that would be accounted for within this percentage as well. And on the right-hand side is the amount of traffic that we have not served. So there is an extremely small amount of traffic that Silverchair will outright block.
ROBB BURGESS: And that's going to be malicious traffic. So someone trying to perform a SQL injection or cross-site scripting. And we're just going to absolutely block that you can't come in. The vast majority of this traffic that we're showing is not served is some sort of automated system, say a bot that has not been approved or a crawler that has not been approved or some other automated system that attempted to go to a site but then could not pass the managed challenge.
ROBB BURGESS: I'm sure everyone's seen that Cloudflare pop-up. You go to a website, you get the little spinny thing, Cloudflare says, checking that you're a human, and then it goes away, and the site pops up for you. That's the vast majority of the traffic that we're not serving is just automated systems that aren't going to pass that kind of human check. So, yeah. And getting into the bot types, as I mentioned, most of the bots in that middle column that we are serving are search bots.
ROBB BURGESS: Your traditional Google bot, Bing bot, Apple bot, things of that nature that's indexing sites to power search around the globe. Some of the newer search bots that we're seeing are for these AI platforms, but they act exactly the same that Google bot does. So you have, say, OpenAI, they have their own AI search bot. And it's doing the exact same thing that Google bots doing.
ROBB BURGESS: It's indexing the publicly available content or web pages on sites, so that when someone using OpenAI or ChatGPT, it's just got a list of-- it's got an index. So it's not having to go search on its own. When you search for something in ChatGPT, it's got its own kind of search index there for you, just like Google does. There are the training bots or the AI models that go train their large language models.
ROBB BURGESS: We block those outright for all client sites and all traffic. We recommend that you keep doing that. You don't want these LLMs training on your content, for the biggest reason of that you lose control of the content. Like, I know it might be rare, but if you release a correction to some article, you don't know what the LLM has slurped up. If it slurped up the previous one, that is incorrect, is it going to pick up the new version that Is correct?
ROBB BURGESS: Maybe, maybe not. But we just recommend that you keep blocking these. And then the user bots. This has been where some of our clients have been experimenting. So this is where, say, I'm using ChatGPT and I search for issues around knee joints. And it goes to one of our client sites and pulls information back from it. That's just going to give me information about that article, say.
ROBB BURGESS: And then it will also give me a link to say I got this content from xyzpublisher.com/kneejoints, whatever it may be. And so that's where we're seeing a lot of the referral traffic come in. So traditionally, Google Scholar has been at the top of the referral charts. And we're seeing the ChatGPT user referral creeping up to be comparable.
ROBB BURGESS: In some cases, it's now the sixth, fifth, fourth highest referral we're seeing. And so this is, as we say, the strategic decision. How do you want your content to be disseminated? Do you want to experiment with AI bots? This would be the place where you would do it with these user bots.
HANNAH HECKNER SWAIN: Awesome. Thanks so much, Robb. So I think it could be helpful to start our discussion today with the publisher in the room, Paul. So as the JAMA Network looks into these sorts of experiments and you all look at the usage from humans and not humans coming to your platform, what are your big focuses in these meetings and what have surprises been looking at your non-human and human traffic?
PAUL GEE: Well, the surprises, I guess, have always been the same. Like we always-- I mean, for 15 years, we're always having spikes around different articles. And sometimes they're human, and sometimes they're not. Sometimes they're from geolocations that cause suspicion and they're human. And sometimes they're bots. Taking a macro view, I feel those investigations are going to continue and always happen.
PAUL GEE: When we're talking about the bots, the thing that is always a balance is we want to be able to serve the content, not just for the business model, but for the mission. JAMA should be something that people have access to is a general idea. And that chart that shows a 53% to 55% turnout rate, really is alarming to a publisher who's trying their hardest to disseminate.
PAUL GEE: That's our job on the publishing side when we're not in editorial. So I think that was the most surprising thing that I've heard and that we've discussed. And when I brought that back to the publishing group that we're turning away 50% of the traffic, everyone was like, well, then there's definitely a margin of error in there that we can benefit from looking closer at.
HANNAH HECKNER SWAIN: Got it. So do you think that, when you decide to selectively allow certain AI traffic in, what were the main tests? Or was it OK, we're blocking these 53, let's be more permissive and then winnow down? Like how did you all make those initial decisions?
PAUL GEE: We followed the advice of people at Silverchair and also people at an SEO vendor. And we looked up and read the definitions of those bots, those user agents, and thought about what they really do. They have a search agent and a user agent. The search agent performs for GPTs, the same thing that Google bot does. It's keeping a wide cache of everything that's available on a daily basis.
PAUL GEE: And then the user agent can be served quicker, the right content. That's, to me, no different than a Google link. And so we kind of took that approach that if our job is dissemination and discovery and allowing aggregators that honor our business models to play that we don't really-- that we actually have a pretty good case to start testing, letting them in. And when we did that, we looked for signs with Silverchair and in our own analytics of any different levels of exploitation than are there already and didn't really see them.
PAUL GEE: We know that we benefited a great deal from Cloudflare when we installed it, because it did eliminate a lot of bot traffic. And when we released this again six months later and started letting partitions in, we didn't see that same return. And we do see healthy levels of page views per session from the sources that are GPT, that are the equivalent of Google and other preferred sources.
PAUL GEE: HANNAH HECKNER
SWAIN: That's great.
SWAIN: Now, Lou, you work directly with publishers. And as a consultant, what experience have you had with guiding publishers exploring this process about where they find the right spot between being too restrictive or too permissive?
LOU PECK: Yeah. It's a real worrying concern, and it depends on the type of publisher. So if they're a society publisher, not-for-profit, commercial, how small they are, how big they are, university, press, and what resources and knowledge that there is, we tend to find that there's a real knowledge to action gap. So we know that AI is here. We know that it's been around since the dawn of computing. We're just a lot more aware about it now.
LOU PECK: And we're now just fully grasping the capabilities that AI has. And so we're finding that some publishers are really paralyzed by scraper's dilemma. And what does it really mean? And what content should we have, and how is that discoverable? And what we let in and what shouldn't? I'm going to talk about some of the things later, but certainly in terms of trust and credibility, that's incredibly important as we move forward in what we do and how different search engines programs and services view how we appear online.
LOU PECK: Google alone last-- I think it was last week or the week before, launched six new tools. And in terms of the user experience and the things that those different tools can do now, it's quite scary. And Google Scholar has a lot that's going to be coming, too, in terms of this is it's year turning from traditional search to more semantic way of doing things.
LOU PECK: So it's very much dependent on the organization, what their core goals are. But really, publishers do an incredible job of ensuring that there's credible research out there. There are workflows and processes really to ensure that what we're publishing is trusted. And there's a lot that goes on with research integrity. But then after that, you've got to make sure that people read it and they discover it.
HANNAH HECKNER SWAIN: Yeah. Those really important core goals of discoverability, reputability, maintaining value. Are there any guiding principles or success stories that you have from working with publishers, as they consider this, as they perhaps open new value streams with these decisions? I know we're still early days with commercializing this kind of traffic, but any sort of experiments or guardrails that you have established in your working with publishers or thoughts?
LOU PECK: Yeah. There's certainly a lot in terms of brand dilution and content devaluation. So there's concern that bots are going to ingest high-value content, and then they're going to mix it with some of their hallucinations, AI hallucinations. And so it's going to make it lower quality. And so in terms of what we're showing readers in terms of what's truth and what's noise, and that's definitely going to be harder.
LOU PECK: What we're seeing that there's a lot of diversification in our industry and how societies or how publishers can generate other types of revenue. And so we're seeing a real movement in terms of advertising on platforms as well and embedding third party partners who can help them to do that. So in some respects, there's going to be different types of traffic that people are going to be wanting to come in.
LOU PECK: So we focus too much sometimes on quantity versus quality. And actually, I think I would rather use an AI tool to do a thousand searches-- to do searches for a thousand scientists, rather than sending a thousand people who are Joe Bloggs, some layperson, to my page to look at stuff, because that really inflates the statistics and what we're looking at.
HANNAH HECKNER SWAIN: Yeah. Thinking about that quality metric, it can be really hard to dig into that. You're not always sitting next to someone who's looking for and finding your content and reading it. So maybe I'll start with you, Paul. And Lou, feel free to jump in as well. Like, how have you dug into the user experience success of looking at this data that's coming in and how can you discern a good experience is happening?
HANNAH HECKNER SWAIN: You mentioned the click-throughs and time on site that you're seeing from the user agents, but is there more deeper dives that you're looking at or additional data points?
PAUL GEE: Honestly, not right now. Other than, does it fit the mix of real human traffic? And if so, should we let it in? Where we're focusing a lot of our attention is building out tools and services for those that do want to use applications. So we're pretty progressive, while we're also thinking very heavily about security. And we keep those in a balance.
PAUL GEE: But I feel like that's the same-- that's the practice that we've had with Google or with other search agents for years, that healthy dilemma of do you work with Google Scholar like 10, 15 years ago, and how can you not? And you end up in a situation where you're collaborators with these companies that are enhancing discovery. So I see institutional traffic coming through these channels.
PAUL GEE: I see high usage and high propensity to stay on the site. So with those three things, it's the fingerprint of what you want to see. And the more qualified to lose point, the more qualified-- if what these tools do is qualify your traffic for you so that it's set up to deliver an experience that's better for the user. And then they're more engaged, they're more likely to site and the traffic's high, then we like-- it feels like a win-win.
PAUL GEE:
HANNAH HECKNER SWAIN: Now we've talked a lot about value, communication, keeping your brand strong, making your content discoverable as decision points in this journey. Robb, do you want to maybe speak about the backend stuff that publishers need to be considering when stuff like this comes up, with regards to infrastructure and things like that?
ROBB BURGESS: So there is a discussion to be had when a publisher initially wants to start experimenting with allowing these AI bots in. For some publishers, it's been a slight increase in traffic, 5% to 10% maybe, which is entirely absorbed within the current infrastructure that Silverchair holds for any given client site. For some publishers, we have seen between 25% and 50% traffic increase just from these AI bots trying to scrape-- not scrape the site, but access content on the site.
ROBB BURGESS: And when it is that level of a traffic jump, we do have to have discussions about protecting the current infrastructure, whether we want to work with a publisher, to talk about a project to expand it. We've put rate limiters on this type of traffic for some publishers that have had a large traffic increase from it. So that's kind of on the infrastructure side.
ROBB BURGESS: And then looking forward, we're also exploring the possibilities of serving this AI traffic in a manner that a machine likes to read. Right now, if you go to any of the publisher sites, go to an article. It is designed in such a way that it is readable and digestible for most of us. And it's not that readable. It is readable and digestible for machines, but it is not optimized for machines.
ROBB BURGESS: And so just removing all of the CSS and all of how it's designed, and just having the content in somewhat of an XML markup format, serving these AI bots just that information is something we're exploring. It's going to lessen the load on our physical server infrastructure and give this information in a manner to these machines that like to read it that way.
HANNAH HECKNER SWAIN: Great. Super helpful. There's some questions that have come through from the chat. One of them is touching on one of the slides that we showed at the beginning in regards to allowing search bots in. There was a note that says block these and you disappear from search results. Should that have perhaps said block these and you disappear from chatbot responses?
HANNAH HECKNER SWAIN: I think the answer to this is no. It's really just allowing the search bots in keeps your content indexed so that it's easier to pick from. Is that a correct statement, Robb, first?
ROBB BURGESS: It's somewhat nuanced. That statement is true. If you blocked a Google bot, you're going to be removed from the Google index. If you search for your site on google.com, it's not going to come up. Now it's somewhat of a different scenario with these AI search bots. For example, ChatGPT or OpenAI, if you ask it a question, and it has to search around the internet for something, it's going to use Microsoft Bing search just like anyone can type up bing.com and search for something.
ROBB BURGESS: And it's going to be a bit slower. It's going to burn more of your tokens if you're a paid user of ChatGPT. But if you have allowed that OAI search bot to index your site, instead of using Bing to search the internet just as is, it's going to use that searchable index. So it's going to be faster and in theory, give the user better results. HANNAH HECKNER
SWAIN: Thanks, Robb.
SWAIN: A recent question that just came through, which I think is hewing back to just the vocabulary that is part of this overall conversation, and it's about the difference between a bot and a crawler. A crawler is a type of bot.
ROBB BURGESS: Correct. So, yes. Not all bots are crawlers. All crawlers are bots.
HANNAH HECKNER SWAIN: A square-rectangle situation.
ROBB BURGESS: Generally, we refer to a crawler if it is something that is trying to consume all of the content. It's programmatically going through all of the pages step by step to consume everything.
HANNAH HECKNER SWAIN: Thank you. Another question that's come in. Not all AI tools provide a link with their responses to users. What kind of ongoing research, the questioner asks, is Silverchair doing-- I think this is also a publisher-- question regarding evaluating the referral value of each bot? What guardrails exist for bots that do not provide robust click-through rates? So I think maybe breaking this up into questions.
HANNAH HECKNER SWAIN: The AI citation counts. Lou, Paul, how have your organizations or the organizations you're working with, how are you all looking into those AI citations? And how does the use of citations in answers-- how does that come into your decision-making when choosing which bots to use?
PAUL GEE: So I don't think we know enough yet. If we're talking about traffic, Right now, we're only allowing in a subset of the GPTs that have a good reputation, and we're allowing the search and the user agents to index and then therefore serve our results. And when we look at that traffic, it looks healthy, like I was I've been saying.
PAUL GEE: How it's used and whether it impacts if we're talking about bibliometrics and citations, we won't know for a couple of years. I think one strategy here is to ensure that we don't take a dip in citations because we're aggressively blocking. And it's a balance of trust. I will say in the beginning, like in 2024, we were probably the first publisher to do a full hard stop robots block for every known AI crawler.
PAUL GEE: We did a wild card. And we did not release it until there was information that we could control it and identify which bots we were letting in, and then we only permit it in three or four. A few of those drive, and one of those really, ChatGPT, drives the lion's share of the traffic. And I believe, I mean, I'm guessing that certain providers like Claude, I don't believe it's because we're not there.
PAUL GEE: I believe it's because they probably have a business model where they want publishers to license their content directly through different channels, which is a growing model that that's coming up where they are starting to pay to play. And we'd be looking at that.
ROBB BURGESS: Yes, I can confirm with all of the publishers that have experimented with letting in these AI bots, 99% of it is ChatGPT. Less than 1% is Perplexity, and less than 1% is Claude.
HANNAH HECKNER SWAIN: Yeah. ChatGPT just sort of became I'll google this, I'll ChatGPT this, that market share thing. Lou, you were going off of mute.
LOU PECK: Yeah. Doesn't roll off the tongue as easy as I'll google that. Doesn't mean that I'll ChatGPT it. I mean, I get confused about acronym all the time. Yeah, I was just thinking. So in terms of how we're measuring things, that's really important. We have traditionally measured things in certain ways and we need to rethink that. So when we think about a bot comes in and it crawls a site and it's doing what it needs to do, it takes like half a second whatever it's doing.
LOU PECK: But because there's no continue click, it's not being registered. So it's registered as a zero click. So in terms of how we're measuring things, and so rather than looking at things like keyword rank, we could be looking at citation share as well. There are different metrics that maybe we as a community need to look at. And I think that NISO and-- who is it?
LOU PECK: Let me just quickly check. I think that there's something going on between NISO and COUNTER, where they're looking at a metric called Access_Method agent. And so really being able to differentiate between human usage and then what's coming in otherwise is going to be incredibly important. What I worry about is I've been doing some research this week speaking to libraries and is the problems that they're facing, which they face every year, which is like a 10% decrease in collections budget or overarching budget for their library, and how that's going to affect publishers and service providers moving forward with subscriptions or different open access models.
LOU PECK: They're pulling apart transitional and transformative agreements. So I'm concerned about the value that they see in the content that they're given in the way that it's currently being given and what that actually means. And part of the way that they measure, there are two primary ways of measuring is cost, but also usage. And so how they measure that usage, is that still relevant?
LOU PECK: Do we need to have a complete change in how things are measured and reported? And that's what I worry about for the future with publishing.
HANNAH HECKNER SWAIN: Yes. Those new counter standards can't come quick--
LOU PECK: Yeah, absolutely. And just in terms of what's been explored and who's using things. And then when people come to your site, you don't want Joe Bloggs just doing a search and I don't know, I've got a mole on my back, and it ends up pulling up some credible research data. But you want the researcher who's actually looking for their research to then come into your site and then experience the content. And that's where we talk about quality versus quantity.
LOU PECK: It is about the quality. But that would mean less quantity coming in. So tricky, eh?
PAUL GEE: I would say I think that we're going to see more and more applications and businesses that are within our space. I think that's only begun. I mean, you can see we've been working with OpenEvidence. That's an example of a very precise market opportunity that's in our area. But I only see more of those growing or being launched. LLM technologies and these technologies are becoming more easy to launch, and they're going to grow up in all of our spaces.
PAUL GEE: Probably medical, it will happen first, in a lot of ways. But it's going to hit all areas of STM. And I think that will be a growing business line for most groups as you begin to license content in new ways, but also you get significant traffic back when they create link back services correctly. One thing that I've noticed, not all GPTs present the links backs the same.
PAUL GEE: Some do it very, very well, like OE and GPT do it pretty well, but others they create layers, where you have to click. And then you get a slide out that then tells you about the thing you clicked on. And then somewhere in there, there's a link, like they're really making it hard to leave that environment, which I tend to think will get them dinged with users. Over time, the users always win.
PAUL GEE: If you want an article, you're just going to either stop using the interface and go somewhere else, or they're going to change the interface to get you to the article.
HANNAH HECKNER SWAIN: Great. OK. We're getting some questions in. Trying to think of the best one to discuss next. OK, I'm going to combine two questions. Just a two-parter directed initially at YOU Robb. When we look at referrals, is Gemini traffic separable from Google Search traffic? So that's the first part. I imagine that will hopefully be a short question.
HANNAH HECKNER SWAIN: Are we able to tell the difference?
PAUL GEE: We are not. Generally, Gemini is going to use all Google bot indexed information.
HANNAH HECKNER SWAIN: Got it. And then in thinking about the use of agentic browsers, how do these fit into the conversation?
ROBB BURGESS: This is a bit of a curveball, and we are not sure what that's going to look like. So I have been experimenting with the claude version of-- it installs right on top of Chrome. And I can go to any website that I have access to. Log in and ask Claude to summarize the data, do whatever it wants to with that data. And from a infrastructure web server point of view, there's no difference between me as a person using Chrome and going to xyz.com, and me using Claude within Chrome to go to xyz.com. There's no difference as far as the server logs can tell you.
ROBB BURGESS: The only way you could do that is if you told Claude to say, scrape a site, and it's clicking through different URLs on a page quicker than it is possible for a human. But other than that type of heuristic analytics, you can't tell the difference.
HANNAH HECKNER SWAIN: And while we have you, another sort of vocabulary question. If a piece of software focuses on just a single site to extract all of its data, is that still considered a crawler? Or should it be classified as a site-specific bot?
ROBB BURGESS: I would consider it a crawler. And if it's approved, say, Bing bot, Google bot, great. I mean, we want to be good partners with all of our clients. And if you tell us you want X, Y, Z crawler to access your site and grab all the information, we're going to do our best to facilitate that for you.
HANNAH HECKNER SWAIN: Great. Thanks, Robb. Paul, this is initially directed towards you from Allison Bellin at Duke. So thinking about specialized AI and chatbot applications, they're proliferating, built to support scholarly research and specific disciplines. So when we think about these bots accessing your content, do they manifest in-the-traffic data? Or are these-- as drama experiences, are these separate deals where it's licensing that you then get usage from that?
HANNAH HECKNER SWAIN: What does this usage outlook look like as JAMA explores these? The way that I'm talking about it here is usage of the site due to referrals back from these discovery portals, I'm thinking, especially in the context of how I've been answering the questions, I'm thinking about them as discovery tools that bring traffic back. And we are interested in getting usage statistics and figuring out how to handle that from off-site aggregators.
HANNAH HECKNER SWAIN: And I think part of that is with a group like ResearchGate, you can get counter integrations back and there's a model for that. We're really waiting to lose point for counter to catch up with this type of solutioning. And then I tend to think we'll have-- we're starting to get all the tools we need in order to do our traditional dissemination work just with this new type of technology, which I think, at the end of the day, Robb, when you saying agentic browsers, regular browsers look the same or feel the same, I wonder how much like this becomes noise in the end.
HANNAH HECKNER SWAIN: We're simply just delivering through technology the content that we create.
HANNAH HECKNER SWAIN: Yeah.
ROBB BURGESS: I think that's absolutely right. You're going to be delivering the content in the manner that you've specified, whether that's the general public going to ChatGPT and trying to access your content that's in front of the paywall, or you have a researcher at some institution that has a license for your content and are using the browser, the agentic plugins. It's basically the same thing. You're disseminating the content in the manner that you've specified to the public, whether that's in front or behind the paywall.
HANNAH HECKNER SWAIN: As we look at that disintermediated future and organizations like the AMA wanting to serve their users' needs where they have them, how do conversations about the JAMA brand-- what does that look like in that future? There's obviously editorial rigor, but if you're not holding up a print magazine, how do you communicate that brand?
PAUL GEE: That's what we're all learning. I mean, not extending your brand, I think, is a problem. And it would be a problem for anyone. If your brand isn't known in the future through some of these tools, where will it be known? Print was the way. In many ways, people have said, we don't read print for years. But print was this pamphlet almost or this reminder that the journal exists, so that when you saw the result in Google, you recognized it and you clicked.
PAUL GEE: And that glue, that intangible glue, it comes up in customer research, it comes up in all these ways. It's not deterministic. You can't analyze it in the data, but it's there. We're humans and we live by recognizing brands. So I think we need to continue to have rigor around the customer research we do and the evaluations we do of how people recognize us over time.
PAUL GEE: I think we all need to as publishers. And make adjustments as we go. We need to be mutable and change day to day, which is how, I think, this game gets played.
HANNAH HECKNER SWAIN: Great. Great points. Lou, I'm going to start with you on this question. It's from Louise Russell. Do you think there will be a shift in funding from library budgets to the institution's technology budget for AI licensing deals? Just thinking this could have implications for how we and our customers measure value.
LOU PECK: Well, sadly, as a librarian, I can't talk so much to this. But certainly, that's a great question. Thank you, Louise. Yeah, it's interesting. I think that-- well, I think the losing budget anyway. So shifting money that isn't there anymore is going to be practically impossible. Shifting money into innovative ways that they're looking at how they can license content, and maybe they might do things like per-article views, things like that, actually only just going for the articles that people are really interested in, which is going to completely causes problems when we look at usage data across the board.
LOU PECK: And because people are only going to be having access to things that they need rather than us bundling things together. So libraries are definitely doing a lot in terms of their use of AI, how they're using it for their user experience, how they're using it in their technology. We've seen that there are universities that have been created in the Middle East, which are very AI-specific.
LOU PECK: That's what they've been created for. So there is a huge amount of effort and there is change happening. But for libraries, it's about, how can they get the content that they need in the best way possible, for the best price that they perceive as value? And that value point is really important. And how they value our content is quite subjective. And a lot of them measure our content in very specific ways, very traditional ways that they have been doing it for years and will continue to do so.
LOU PECK: So that's potentially problematic. But to say to Louise's question, yes, potentially. But I think there's lots of things that they're looking to do with their money that has been coming historically to us in this industry. HANNAH HECKNER
SWAIN: Another bit
SWAIN: of a crystal ball question from Stuart Maxwell. Could there be future cases for researchers to evidence that they have actually accessed underlying content, rather than just relying on agentic outputs? So you can't cite Wikipedia in your paper argument here. What might that look like in the future?
PAUL GEE: That's what DOIs do.
HANNAH HECKNER SWAIN: Yeah.
PAUL GEE: I don't know if there's-- I like the question because it made me think like about if there's a way to fingerprint these and use that. And I mean, if there's a technology that allowed citations to be validated whenever or wherever they went, then that would be used, I'm sure, by all the researchers who want to be able to trust their sources. I think trust goes two ways.
PAUL GEE: I can't imagine that researchers feel a lot of trust. They're excited that they can get information quickly. But how do you validate what you're seeing with these new discovery tools? But it's also compounded. I mean, the amount of content that's published in the last 25 years, year on year, the increase is exponential. At a certain point, we simply need to make discovery easier.
PAUL GEE: And these tools do that. So finding ways to pinpoint authority and marks of integrity and trust are equally important.
HANNAH HECKNER SWAIN: Yeah. And this brings to mind a conversation that we had in one of our prep calls about how growth in AI usage could actually be a huge democratizing factor for access and research. Lou, you had some good points on that. Mind sharing a bit with the group?
LOU PECK: Yeah. No problem. So one of the things that I just want to mention is that AI is the way it looks at brand, because we were talking about brand, is that it sees brand as a fact, whereas we see it more as a value. Guys, sorry. Sorry, my kids have just come in honestly. And so we think about it as a logo, an experience, what's it look like, what's it feel like.
LOU PECK: And it just doesn't look at us like that. And so it looks to a brand as an authoritative answer. And it's about just what Paul was saying, what we said before about trust. And so when we think about what's coming into our site, what's linking to our sites, what's linking to our content, all those spam, all those naughty sites that are going, trying to be credible themselves, building up their trust score, their Google trust score.
LOU PECK: You can actually use a Google tool to reject them, to actually cut them off. So a lot of the time, we're building things with our eyes. And so we really need to be building things for intelligence. We optimizing for a bot is actually optimizing for clarity and accessibility. And we know how incredibly important that is. So if a machine can't understand our research hierarchy or our metadata, then actually, how can someone like myself, who's neurodiverse or anybody, really, how can they using assistive technology if they've got a disability or differing ability?
LOU PECK: They probably can't either. So we need to really be thinking about the future of what we're doing, how we're presenting data. Is it in other ways rather than just in a PDF or HTML? Is it in a video? Is it in audio? How are we disseminating that important information? It's much easier to present a video of an experiment than to write about it.
LOU PECK: So it isn't just about links anymore. It's about how a map of research relates. And so our platforms need to be bilingual. So you've got this beautiful layer, the Silverchair, and underneath all the other stuff that Silverchair is doing. But it's about what's engaging for a human and then also the AI that they're using. So it needs to be clean, structured, machine-legible, but also, for us, ourselves, about how accessible is that content.
LOU PECK: And we need to be thinking about not being where we want users to be, but we need to be where users are, because we can't force people to do the things that we expect them to do.
HANNAH HECKNER SWAIN: Yeah, definitely. Great. Let's see. We'll have two fun questions to close this out, and I'll open the poll after this first question. So as evidenced by this conversation, even though people that think about this all the time, we're doing lots of experiments, we're learning a lot, but we're still doing things in real time. So with this fast-moving environment, what do members of this group wish they had known a year ago that-- like things now that you wish you knew a year ago, and what are you trying to figure out?
HANNAH HECKNER SWAIN: Robb, I'll start with you.
ROBB BURGESS: Seems like a loaded question. No. I wish I had dug deeper into the underlying infrastructure of the large language model, the RAG retrieval models and how that all works. I think I would have been a bit further along in the journey than I currently am.
HANNAH HECKNER SWAIN: Anything we're still trying to figure out? I think a lot. What's your number one thing?
ROBB BURGESS: Just how to use it effectively. But there is always more learn. And using it from both sides, maybe I don't know if offensively and defensively is the correct term, but on one sense, from part of my role is to protect the Silverchair platform. So being defensive on that end from unapproved AI bots, crawlers, all those good things, but then also being on the side of what we need to employ it to do better, to do things faster, to make sure that we are moving along with the times, that we're keeping up to the cutting edge within our platform and for all of our clients.
ROBB BURGESS: HANNAH HECKNER
SWAIN: Thanks, Robb.
SWAIN: OK, Paul, what do you wish you had known a year ago?
PAUL GEE: Well, what I wish I knew better now, and should have started a year ago, is just understanding how to ensure that as we allow this traffic in that we're getting it to the right content. And I was thinking when Lou was talking about-- something you said made me think about the value of SEO. Like when you initially started doing SEO, everyone would say, well, you're writing that copy for the bots, like a blurb at the top of a page that explains what the page is for.
PAUL GEE: It's actually for humans. It makes your sites better. If someone can actually understand the context before they dive down to the bottom. And I see us starting to go down the same routes. If we end up making special types of metadata around our articles or interfaces that people don't see, but the bots consume, it stresses me out a little bit that we're making better content for the bots than we are for the humans.
PAUL GEE: As publishers of primary literature, we prefer the primary literature, but humans really do enjoy context. And I wish you learn these things real time, but as I see that coming up, I think that's a really fun territory where Publishers content could get better not just under the hood as they try to feed the bots better sources, but using it within the interfaces to make the experience better as well.
HANNAH HECKNER SWAIN: Yeah. Where the search engines optimizing content, that was written best for human consumption. And are the bots doing the same? So we should just human optimization, it becomes all one thing. That's great. Lou, what do you wish you had known a year ago?
PAUL GEE: That clicks would become a vanity metric. So we've always been thinking-- well, not always. We have recently been thinking of AI as more of a discovery tool. So that leads people to sites and they can view the full paper. But actually, it's more about they don't want the paper, they want the insight, which goes to what Paul was saying. And so search referrals, we've seen them plummet by about 33% to 55% in the last year.
PAUL GEE: So I think it's really better understanding that metrics are going to change. And actually, it's not about optimizing for traffic. That's what we've traditionally done traffic, but start optimizing for inference interference present. So ensuring that our content is the preferred source for RAG, that it's trusted, that it's credible, and that we continue to really support with everything that we do already for that content, that we continue to support that in a way that it remains discoverable with so much noise, because there is an incredible amount of noise out there.
HANNAH HECKNER SWAIN: Great. Thanks. OK, Steph, I think we're ready for this final poll. And as folks are filling this out, so how has this session changed your thinking on bot access to content? Indicate your thoughts here. While everyone's filling that out, we've got a robust group of folks attending this call. So I'm going to go around and ask all of our panelists to say, one thing publishers should stop doing when thinking about AI traffic and one thing they should start.
HANNAH HECKNER SWAIN: Robb, I'll start with you.
ROBB BURGESS: When thinking about AI traffic, again, we've said this in every one of these webinars. I think you have to go back to your organization's stated goals. Is your goal to disseminate this information to anyone, everyone, all the time, everywhere, no matter what? Or are you more of the. We want to make sure that this information is as accurate as possible and serve to our authorized users.
ROBB BURGESS: There's a difference in those two statements. So I think it's going back to your mission statement and really deciding what your organizational goals are.
HANNAH HECKNER SWAIN: How about Paul, things that publishers should stop or start doing when thinking about AI traffic?
PAUL GEE: I guess. Stop having an automatic bias against bots. I see a lot of the debate about humans being preferred and bots having a lot of negative attributes. I tend to think that the humans gave the bots a lot of negative attributes, and this is it's not so easy. Humans and bots are very similar in the way that they tend.
PAUL GEE: to be communicated with and the way that we need to work through them. And I see that metaphor in a lot of different ways. But I think not trusting the bots to the point of treating them they just need to always stay outside the fence is only going to hurt the longer and longer we take that position. I think that if our job is dissemination, we need to figure out how to disseminate with these new technologies and to not fear them as much.
HANNAH HECKNER SWAIN: Great. OK, Lou, bring us home.
LOU PECK: Yeah, mine's quite similar, actually. It's about stop treating AI traffic as noise and something to be blocked. So you can't-- the future's here. The future's now. It's coming. Well, it's already here. So you're only actually going to end up making your research invisible. So start focusing on generative engine optimization.
LOU PECK: Paul talked about SEO. We love our acronyms. There's a GEO. But this is about ensuring your brand, your content, your journal, your books, whatever it may be that they are trust anchors because that's what AI models that they rely on that truth aspect. So a website is no longer a final destination. A content platform is no longer that final destination.
LOU PECK: It's more about verifying. So we need-- users need to discover the insights they'll discover in an AI interface. But when they come to you as a publisher, you need to. You need to show that it's real. So make sure that you're ready for them. HANNAH HECKNER
SWAIN: great love it.
SWAIN: Keep doing the things that have made you a respected brand. Be open to experimentation and remember your mission. Great things to end on and could be applicable to anything. So yeah, what does this poll say? OK, more open to experimenting. That's great. Wonderful. Well, Robb, Paul, Lou, thank you all so much for your insights and the conversation today. Stephanie, thank you, as always, for your wonderful organization.
SWAIN: Any housekeeping notes for us to end on?
STEPHANI LOVEGROVE HANSEN: Nope. All set. Thank you all for joining. We will send the recording around afterwards and really appreciate all your engagement throughout.
HANNAH HECKNER SWAIN: Yes, thanks everyone.
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