Reviewing Our Data Trends & Predictions of 2025 with DataCamp's CEO & COO, Jonathan Cornelissen & Martijn Theuwissen

Generative AI is transforming industries at an unprecedented pace. But as AI changes how you or.

Your team work, one thing is clear.

Your skills also need to evolve.

At datacamp, we offer everything you or.

Your team need to adapt and thrive with AI. Whether it's business users looking to get the Most out of ChatGPT and Copilot, or developers and data scientists looking to fine tune models, you can learn the entire AI skills spectrum on Datacamp.

Power your AI transformation today.

Start learning@datacamp.com. Welcome to Data Framed. This is Richie. It's a new year, which means it's time for another round of predictions about what's going to come in the following year. In this first of two sessions, we're going to review the trends and predictions that we made last year. I'm joined by datacamp's co founders. Jonathan Cornelissen is datacamp CEO and Martin Thoison is datacamp coo. Let's see how well their predictions fared. Hi there, Jonathan. Hi, Martin. Happy New Year. Welcome back.

Thanks for having us, Richie. Happy New Year.

Happy New Year, Richie.

Yeah, I'm quite excited to get to grade you on your predictions from last year. It's a good excuse to be harsh with you both, but before we get into that, I'd like to note, like, what have you been most excited about over the last year in the data draw AI space?

I think my answer is quite high level. But just overall, all the progress that we've seen, it was far from a boring year. Felt like every week, maybe every day, there was an AI announcement that was more shocking than the AI announcement. Before we had tons of new models from existing players, from new players, open source, closed source. There is a little bit of drama as well in the AI world. Lots of people who are openly fighting definitely across the foundational models. Definitely wasn't a boring year ahead of.

Technological progress, but also a lot of gossip going on and yeah, a bit of drama's always good to spice things up.

What's fascinating is that things have not slowed down. And that's exciting, right? Because I think a half a year ago there was some talk around are we kind of nearing the end of the S curve? It seems like the answer is so far is definitely no, we're not at the end of the S curve. This is probably just the beginning.

That's true. I think last year we were talking a bit about have we reached the point where scaling no longer works and models are good enough and AI is basically done? But yeah, there has been a steady stream of progress still. So that's been pretty exciting. Let's get on to the review of last year. So your first prediction and read this one out, was that the OpenAI and Google duopoly at the top of the Chatbot arena leaderboard will be broken. If you cast your mind back 12 months ago, there were basically like it was just neck and neck between OpenAI and Google. Like they would each release a model and it would top the sort of all the benchmarks and they would go backwards and forwards. And that was basically, that was it. So the question is, has that changed or not? What do you think, a duopoly there or do you think there have been some other challenges?

I would say this is more or less correct. If you purely look at the Chatbot arena leaderboard and we can put the links in the show notes, it's run by Berkeley, some researchers at Berkeley to be clear. And if you look at what the top 10 is today, so Google is still high up there, but in the top 10 you can find the OPUS model from Anthropic, you can find Grok and you can actually find Both above the GPT5 model by OpenAI. I would say if you look at that ranking, it's yes. Now I think the caveat is that the strength of the models is not reflected in the market share of the models. If you look at, okay, what are probably still the most talked about models, the most used models, you're still stuck, quote unquote with Gemini and OpenAI's model. That said, I think Anthropic is a steady riser. Uh, I anecdotally hear and see it a lot more definitely in the coding environment. So. But from, from my perspective, if I would grade it, I would say it's a yes.

That's interesting. Uh, yeah, uh, if we look very closely at the, the terms and conditions on this one, it did say Chatbot arena. And uh, yeah, just looking at this, currently Gemini 3 Pro is number one and uh, it's actually Grok 4.1 thinking is the number two spot. So Grok has certainly sort of broken that duopoly on the Chatbot arena leaderboard. I was looking at Data Web though and yeah, currently saying Chat GPD has a 68% market share and Google's got an 18% market share in the LMM market. So between the two of them, they're still very much taking all the market. Maybe I'll give you half a point for this one since, since it's right in some sense now, one thing I.

Would say is an interesting thing to look at is I think what's, what's not covered in the similar update is actually like how open source is doing. And to be clear, I don't have the answer either, but I actually wonder how many companies run open source or deployments of the open source model. Whether it's Deep Seq, whether it's the Gwen models by Alibaba, it's unknown at the moment. I haven't seen good projections there, but yeah, I, there's probably some hidden traction as well there.

I think that's totally fair. I think the trend is clearly correctly predicted because if you go back a year, it's hard to underestimate how dominant OpenAI and ChatGPT were at the time, with a bit of an influence of Google. And if you look at it today in terms of performance, the proprietary models are much closer. You can debate it depending on the benchmark. And yes, market share, it's mostly Google and OpenAI. But the trend is quite fascinating, which is OpenAI is losing market share kind of on a very consistent basis month to month, even though they're still kind of the juggernaut. Like if you compare that to the early days of search engines, I think once Google established its dominance, it took how long, 20 years for that to kind of start going down. So we're definitely in a different type of market.

That's really interesting actually. Yeah, I suppose there's less of a sort of network effect from Google had more data than everyone once they got big and that just meant that they could get better search results. And maybe there isn't that equivalent sort of effect in the LLM space. Or maybe it's just the competition is just so hot that yeah, all these companies are competing so hard that the leader is bound to change at some point. All right, let's move on to the next prediction. So that was that the new AI reasoning models believed with a scientific breakthrough. This is your prediction, Jonathan. I'll let you grade your homework as well. Do you think we have had scientific breakthrough, those that have been directly from artificial intelligence?

I was on the fence about this one because I feel like a lot of the other predictions hit. This one I think is highly debatable. I think when we made this prediction we sort of positioned it as AI can do this independently, which I don't think has really happened. I can't think of any, any, any examples. What is true, I think is that like the, the productivity of researchers in almost every domain is going through the roof as a result of AI, but that would be cheating because I don't think we intended to predict prediction that way.

I think at this point, if you're a scientific researcher or, well, almost any research, in fact, you are going to be using a form. But yeah, we're yet to see some major scientific breakthroughs just from a bot running as they go, like by themselves. So I think, yeah, we, we were talking about the new, well, the then new deep research models. They were promising to be able to solve really complex technical problems by themselves. And yeah, it's not quite at the point where it's done science by itself. I think this one, it's kind of a miss, I'm afraid.

I think so, unfortunately.

Okay.

All right, we're half a point out to two then. So a slow start, but let's see how we go with the next one. So this next one is the one for you, John. So the synthetic data is going mainstream? Yeah. Do you think that's happened?

I think so, yeah. I think this one actually the trend started probably more than a year ago, but I think by and large the amount of data in the world is limited. And I do think there's more and more use of creative ways to create synthetic data to improve models. So I would say this was a correct prediction. What do you think?

Yeah, so I mean, I find this crazy because when you go back like a few years, there was talk like the amount of data in the world is growing exponentially and it seemed like an almost infinite resource. We've reached a point where, yeah, certainly in the generative AI space, all these frontier models, they've already used up all the data that's publicly available. And so, yeah, they've turned to synthetic data to train the models. I think particularly in finance and healthcare, where you've got the sort of privacy concerns, there's a lot of uses, synthetic data there as well. Just because if you want to publish something, make something public available, you can't make use of the real data. So, yeah, I think it has gone mainstream. I don't think it's mainstream in the sense that the man on the street knows what synthetic data is. But I think certainly within our domain, that one's gone. That's correct. The next prediction, that was one from you, Martin. So this is that a video generative AI finally takes off. What do you reckon?

Yeah, so for that one, I would definitely say, like, it's a full point. So I think at the end of last year we were very much into to like experimental videos were brought out, like short clips of cats Cooking and this kind of things. I think if you now look back 12 months later, you actually see a lot of those AI videos thanks to new models like the new Sora model, the new VEO model and all the hype it created. Like we actually see production ready videos being used. If you look at some of these shorter influencer videos, if you look at social media ads that make use of video, more and more of these are being created with AI. I see that within datacamp, I see that with other organizations. The fact that it's now in production I think is really like the check mark on like this thing took off just like 12 months ago. Like the AI images took off and like the years before that, like AI text or contents took off. Like I think we now really have the AI video taking off and frankly I think it's only going to get better as we will be able to create longer and longer videos and the use cases will grow. Absolutely.

I mean I think for a long time you could really spot when something was AI generated. Like you know, I mean start. It was like the hands and then there's like, if you look at stuff in the background there's always like some signage that was a bit wonky or something like that. So now it's like, oh yeah, you really have to stare carefully at something to see is this AI generated or not.

And it's super accessible. Like just it's prompting like you, you, you type in the text, you tell it like this is what it is supposed to do and it, and it does it. And that accessibility is really helping it in everyone in the creative sector to, to start using it. I wouldn't underestimate the ease of use here in the capability of it taking off.

Absolutely. The, the progress has been tremendous over the last few months in particular. So actually I'm, I'm curious, maybe a question for you, both of you. Are there any particular AI videos that you've seen that have you.

There was, I think that went around like a week or ago, like start of January. Like there was this, it was an unofficial video, was an unofficial advertising for Hermes. Like the bag company. Well, they do more than bags, but that's as far as my knowledge of Hermes goes. Like they, they sell really fancy bags and there is, yeah, basically a whole ad created on that that looked, that looked amazing. Like it showed the product as it is. So they were able to use the actual product, it stick to the brand guidelines it was clearly inspired by previously. And that's all stuff that you normally need to build from scratch and code. It affected the model and you can give that through the prompts. To create something in this style is enormously powerful. And I think actually some of the ads that we created with datacamp were not too bad either. Some of our course announcement videos which actually used your voice. Richie. Not sure if you want to speak a little bit to that on like what's possible here.

Oh yeah. So I spent a lot of time chatting to ElevenLabs. So ElevenLabs is voice AI and we've basically cloned my voice. So the podcast of course will always be human. Richie. But yeah, for shorter advertising clips, I'm okay with my voice being used for those sort of things. It saves me being hassled by. By the marketing team.

Yeah, so that's pretty cool.

Joe, how about yourself? Have you seen any particular videos that you.

There's not one that comes to mind, but I'm a. I'm a heavy consumer of kind of AI generated meme videos for better or worse. Interesting.

So there was a bit of a phase a couple of months ago with like just random bits of scenery being destroyed. I thought that was like a fun sort of trend is like random just AI destruction videos, which was actually. Yeah, one of the things with me is like they tend to be like quite niche at this point. There's like trends happen and maybe not some, maybe you don't. All right, so let's go to onto the next prediction. That was that a generative AI product that takes more than an hour to respond with launch. So we've gone from just trying to get text to respond as quickly as possible to like sometimes you want to wait and get a better answer. Joe, this was yours. Do you want to talk about whether that's happened or not?

This feels like definitely a trend that we correctly kind of predicted. I think some of the new Claude models can think for hours if they need to. I think the honest take though is that the sweet spot for say coding agents is still kind of sub 20 minutes for like platforms like replit. Lovable. I think you can get into the minutes or maybe 10 minutes, but they kind of tend to get lost if they take more time. There's definitely products that do this now and I think humans are adjusting to giving higher level tasks to AI agents, whether that's Claude or replit or lovable, and just having a few of those things run in parallel where they come back after a few minutes. And so this seems like a trend that is continuing. So I would give a full point. I think that brings Us at three and a half out of five so far if I'm not mistaken.

I think you're very focused on the points here.

Absolutely, I'm result oriented.

Sorry, one thing just to on this, what I think is important to reflect on like 20 minutes is a lot of time like this, this whole AI wave has been going what for like three or four years and like we are already at having an AI agent basically doing a task for 20 minutes without any intervention. The high quality I think for everyone who's listening in, that's really powerful. So early in I can imagine a future where a couple of years from now like this is seven hours, this is a full working day that you come into the office and you get your AI agent or AI agents to work and they basically execute for the full working day. Yeah, this, this makes me really excited. Yeah.

It's sort of a double edged sword though because obviously if you're waiting for stuff then and it comes up with a bad answer then it can be very frustrating. I'm wondering how far companies can push this. Would you be willing to give an AI task and they're like wait a month for it to think deeply and come back with something.

But does I find a bit strange? It seems like we evaluate AI at a much stricter condition than we evaluate, for example, employees. You can have the same thing with when you manage somebody, you give them a task in the morning and at the end of the day you see the result and you think oh, this was probably not what I intended it to be. I think yes, that risk is there. Two, I think the market will fix it because like those models that will have that risk, less will win. Yeah, I think we put the bar sometimes too high.

Yeah. To be honest, I have a bit of a different opinion here. In the sense that or a different perception. You anthropomorphize AI to the point where you're like, hey, if a human would do this in 10 minutes, it's great. To me sometimes it feels like dial up Internet where I'm like, I remember the days where you would open like five websites in in parallel and then you would wait and you would go back to the first one because by the time you kind of typed five URLs like the first one would have loaded. I, I, I catch myself expecting an immediate answer. Even if it's a task that would take a human 10 minutes. I, I do, I, I think that the two strengths are going to be there like on the one hand, like they're going to Reason for longer and their, their capabilities are going to increase. But I think there's, there's an enormous amount of commercial pressure to make, to make these things faster. If you just think about our use case of having an AI teacher, an AI tutor deliver the learning, like whenever it's more than a second or two, humans immediately get this perception of like, hey, what's wrong with this teacher? Even though like a real human teacher sometimes may have to think for a few seconds when it's AI, I'm immediately like, hey, what is this thing doing?

Yeah, that's fair.

Absolutely. And it does depend a lot on the use case. So certainly like chatbots, they want to be instantaneous. But yeah, if you're crunching really difficult problems, you like, you want to go back to the idea of AI solving a scientific problem to create some kind of breakthrough. Like, I feel like you're probably like, I'm okay with it taking a bit longer to solve a long standing problem and write a paper. Okay, all right, so let's move on to the next one. So it's a prediction from me, Martin, on saying AI features will stop being labeled as AI except in tech products. So yeah, do you want to explain what this one was about First?

A lot of products basically slapped on new features that they were releasing the term AI, AI for this, AI for that, and so on. And if I would grade my own prediction, I was 100% wrong on this one. I think that it's still being used pretty much everywhere on almost everything that gets released by company Intech or Northern Tech. So AI seems to be, seems still to be a very valuable marketing term. Hype cycle is not dying. And I definitely wouldn't make the same prediction for 2026. I think we're going to be stuck to this for probably a long time.

Yeah, like there's definitely a point where you're like, do I really need AI in my sandwich? But apparently it works as a marketing term. So I guess.

Okay.

I have to say I do think this is a bit counterintuitive to me as well. For what it's worth, people are not kind of tuning out the AI labeling and that it still needs to be there. But, but I think that shows we kind of live in a bubble where that's been around for a very long time. That's my guess. That's why it feels odd to us.

Yeah, I suppose the three of us, we think about AI stuff like on a daily basis, so we see it everywhere, whereas I guess most people are slightly less exposed yeah, we expect it.

Everywhere at this point.

So, yeah, I guess good for marketing people. We'll see how long this continues. But yeah, maybe at some point we'll just go back to just selling stuff rather than saying stuff with AI. The next one, this one from you, Jonathan. So saying the dominant AI conversation for boards and executives will be AI usage. I think this one needs a bit of background explaining.

Obviously this is something that's top of mind for datacamp. We help organizations with their data and AI transformations. And one of the reasons we get brought in is when organizations invest heavily in data and AI infrastructure and then notice that there's a second component to that which is enabling the workforce to use that data and AI infrastructure in a very effective way. And so we've seen this firsthand in the past year. It's a board level conversation, it's a C level conversation. Datacamp has been preaching the importance of data literacy and AI skills for quite a while. And it's, it's honestly been incredibly exciting to see that, that kind of level of attention and traction in the last year accelerating our B2B business very significantly. And I think that's, that's one of the objective singles. I would say that we have to say this is, this is definitely happening. Anything you would add there, Richie, do you agree?

Yeah, I mean, certainly, I mean, used to speak to executives and boards much more than I do, but anytime I've run a webinar that has AI adoption in the title, it's been incredibly pop. I'm curious though, as to whether the other sort of executive level conversations you've been having around AI, like beyond adoption.

And usage, the initial conversation is really about AI fluency. And depending on the organization, that's still a huge focus, basically about ensuring that people understand how to use AI to be more productive as well as understand what the ethical risks are and other types of risks in terms of AI usage. I think for companies that are a little bit ahead of the curve, the conversation is starting to shift towards one layer higher, where it's not just human with an AI copilot, but it's essentially humans designing AI agents and starting to collaborate with AI agents. So that's one area where we're seeing the conversation starting to shift. We're also starting to hear more and more that it's not just about usage. A lot of dollars have been spent in the last two years, so companies are starting to ask the ROI question, where is the return on investments? And then last but not least, I think as AI continues to Change this fast. There's a continuous discussion around AI skills and that is obviously what datacamp is focused on. And it remains very much top of mind because AI is changing so fast, where people, maybe every knowledge worker one year ago had to learn how to use AI as a copilot for various things. Now there's all these other ways they can leverage AI, all these other ways that are creating risk and blockers that need kind of reskilling and upskilling for people in different roles.

That's very true. So certainly once you've got people to use AI, you need to make sure that they can use it well. So that's the upskilling part. And also you need to make sure the business is getting value. So that's the ROI part. And I think, yeah, you need all three parts together in order to like save both. Should we actually do more AI or not? That's how you're going to get to winning. Two more predictions to go. So this next one from Martin. Due to AI, sorry, due to regulation, Europe will fall behind in 2025 to a point that catch it will no longer be possible this decade, both for the development of new AI models and applications and the adoption of AI. So this is a slightly negative prediction. So talking about European regulation, Martin, do you think this has happened?

So as it happened that I think they fell behind even more. In short, I think the answer is yes. If you look at a couple of data points, for example, if you look at the investment that is done in the US compared to Europe, you see that the US is doing 10 times as much investment into AI, its infrastructure as is happening in Europe. Secondly, if you look for example at the foundational models which are pretty much the basis of this all depending on like, okay, who you think is willing, like underneath it there are these foundational models in the eu, you have Mistral, nothing else got added or nothing meaningful. And it seems that we also didn't catch the open source wave. So if you look at, okay, what. What was in 2025 big narrative besides existing foundation models was the release of a couple of newer open source foundation models, Deepseek and its new models, Gwen and Swan. And none of these open source models that made hotwinds came out of Europe. So I think that ship has sailed. And like we are missing the investment and we are missing, I speak we because I live in Europe. We're missing the investment and we are missing the foundational layer where I think there is to. To. To make it slightly positive and say, like, okay, where There's a little bit of hope. It seems like there are a couple of very promising applications being developed. Like we were talking about 11 labs. Like okay, like cloning. Cloning. Richie's voice, which is from Europe. It seems in the, in the vibe coding space that we being Europeans have a very good contender with, with lovable compared to for example some of the US players. So it seems like, okay, we're making some, some happens there. Like there's talks about like the EU AI act being softened. Now I'm from the viewpoint like, like any regulation is too much regulation and let's still see what type of regulation softening they will do. In short, I think the prediction is right. I think if you look at what I think key factors are for national models and investment, like we felt further behind simmers of hope in the application layer, but let's see how that plays out.

Absolutely. So certainly in terms of the amount of investment in AI, yeah, Europe just can't really compete at the moment. I guess Silicon Valley just has such a stronghold and there's like so much cash flowing about that. Yeah, there's no way Europe can beat that. So on the regulatory side of things, I do think things are quite promising. So there's this, you said there's going to be some simplifications, the EU digital harmonization package. So it is going to make the regulatory burden a lot easier around data and AI currently is kind of, it's basically impossible to comply with the EU AI act and GDPR and the EU Data act and all the cybersecurity regulations all at once. So yeah, I think the timeline for this was like a turnaround by 2030. I'm hoping that the EU is going to make it. I'm not sure whether we're going to see any sort of new startups. Jonathan, you wanted to add to that?

Yeah, maybe to add one thing. I think if you zoom out, this seems to now be a race between kind of US and China. And what continues to fascinate me is that China is kind of leading in the open source on the open source side, which I would still bet on open source eventually kind of dominating in, in this, in this space, but I would also not bet against the US ultimately figuring that out. And, and, and in some ways like Europe is, with the exception of some application kind of level startups and, and Mistral, Europe is just not really part of the game. It's, it's really kind of a US versus China game. You can, Yeah, I feel like we, we got the full point here by, by relying on European politicians shooting themselves in the foot.

Perhaps the European politicians were just listening to Martin's advice that less regulation was needed. And so, yeah, it's the start of a turnaround. But yeah, I agree there's like half a dozen like really amazing European AI companies, but it's not at scale of like us or China. All right, I'll be generous to let you have a point for that one. The next prediction, our final one single bond from Martin. So this is in 2025, we'll see that the individual, at the individual level, that proportionally teachers will be the biggest adopters of AI in their day to day work, not business folks. Martin, why do you think teachers were going to be big on AI?

So one, it's accessible to them too. I think from a teaching perspective, using AI, you can do a lot of cool things to make your lessons better, more interactive. So I think if you approach it creatively, like it can be your new blackboard that you do, like the physical thing with chalk, you have different students in front of you. And with AI, you can actually tailor your lesson materials to the type of students you have in front of you too. If you think about the prep work that teachers need to do, like whether it's lesson plans, whether it is creating the necessary presentations and so on and so on, like you can do a lot of that much more effectively. And I would say third is that a lot of the information that teachers teach is actually already available on the Internet. And an AI tool like ChatGPT makes that very relevant, accessible in a very easy format. So you basically have alternatives to your textbooks and so on. So that was a little bit the underlying thought experiment in terms of where we are. It's, it's a bit hard to judge because I don't think there's any, any clean data available. But I did look a couple of things up which seem to indicate that teachers are definitely ahead of the curve or very heavy users. So OpenAI, I believe earlier this year released some data on top use cases for OpenAI. And what you actually see there is that teaching is pretty much up there as a top use case. Secondly, for those listeners who know, for those who don't know, like at datacamp, we run a project called datacamp Classrooms. Basically we give professors and their students free access to the whole Datacam platform for six months as part of the class that they're teaching or that they're attending. Now that means that we have a large group of professors that we can survey and we Surveyed them in the second part of this year on their use of AI in their teaching activities. And what we actually found was that 93% of these educators are saying that they use AI tools and close to 70% are using them daily. And anecdotally, like, those are not the figures that I see when I talk to our business customers. So from that I intervene. Like, okay, like, teachers do seem to be very active on the AI tools. Another interesting thing that came out of that, which, which gives this a bit of a different view is that actually when you ask educators, like, okay, how do you think your school is doing that? They actually say that around, like, okay, only half of these teachers get support, like formal training in how to use AI for their teaching activities. So there seems to be like, no centrally organized, like, okay, we're going to upskill all our professors in how they can make best use of AI in their classes from the university or the school perspective. So I think that's like, that, that's one, is, is a, is a missed opportunity. Two, I think it speaks a lot to the internal motivation, intrinsic motivations that teachers have to put this tool to use and the need to have to put it to use. Like, given we see like over 9 out of 10 professors are, are using it. So I'll let you do the, do the scoring on where this, where this ends up. But for from what I see, like, teaching is a big use case and the educators that we surveyed, they are using it heavily.

AI is incredibly useful for creating educational content. We see that at Datacamp. I mean, I think almost everyone, Data camp is using AI in some form to help teach people. And so I can certainly see the use cases are just almost infinite for the education sector. So it's interesting you're saying that, what was it, 93% of teachers are using some form of AI and like about 2/3 of them using it sort of daily. This is like, it's not far off the sort of levels we see from our state of data and AI literacy for people in the IT sector and data team. So these are generally the sort of the teams that use AI the most. So that's actually, it's not far off business. I agree. Some of the business teams, like the sales and customer success and whatever, they use it a little bit less. So maybe teachers are using it more than a bit more than the business folk.

And I think it's partially because the kind of early adopters are always the youngest people. So you have a ton of students heavily using it. So I think it's also, they have no choice. They have to kind of catch up because the people in the classroom are adopting this and pushing them.

Yeah, you don't want to be outsmarted by a bunch of kids. So, yeah, got to keep up with the children or at least get the good ideas from the kids in your class. And then, yeah, you see what you can use it for. But certainly, yeah, creating lessons or doing research to find out what's useful. These are some great use cases for education. I thought it was last one. I missed one final prediction. You know, by now, like, I got two maths degrees, I should be able to count to 10. Okay, so final prediction. This is another one from Martin. So this is the AI becomes the killer use case for blockchain. Talk us through it. Like, what sort of use cases were you envisaging?

Yeah, I was. The thesis was that because, well, true AI, you can pretty much fake everything. You can make fake riches, you can make fake images and so on, there was going to be a need to actually validate that when you bring out an image, when you bring out a video, that that's actually you. And the blockchain is a useful technique for that. That ID became mainstream. I don't think so. So I think this prediction did not play out. Doesn't mean, like, it's not going to happen. Hard to say when. We're going to need to have some type of, like, actual, like, proof of identity in the. In the future. It's all the capabilities that AI has today and will probably have in the in the future. I think AI swap was this new term invented in the last three months that will create a need for saying like, okay, this is like, there's an actual human being behind this. Or like, or I approve this message. This is not something that was created and then wrongly attributed to me. So miss on the prediction, but we'll need at some point in the future some kind of like proof of identification or proof of identity.

Absolutely. So certainly deepfakes are a big problem there. It's only going to get worse. And so having that kind of provenance, particularly I guess for political messages and things like that, or if you've got some celebrities wanting to be like, this is really me saying something then, yeah, that. That seemed like it's a big problem we've got to solve. Maybe blockchain is going to be helpful there. But yeah, at the moment we've not seen a great deal of action in this area.

Yeah, it's quite interesting. I actually, I expected there to be something in the last year or in the near future. I'm starting to shift my perspective and think, well, maybe humans are just gonna adjust to the point where we don't need this. Because now when I see a video that's somewhat surprising, I just assume it's not real. And usually that's. That's actually the correct assessments. So, I don't know, maybe humans will just adjust and not believe most things they see. It's a fascinating dynamic.

Absolutely. So, I mean, I think about my wife in these situations because she tends to believe almost anything she sees. And that's kind of fun from my point of view sometimes. But in general, I'm like, yeah, I don't know how she's gonna cope with AI everywhere where it's really believable. So, yeah, hopefully people will adjust, and maybe we'll just get to the point where we are better at sussing out, well, is this real or is this not? Hopefully there will be a solution to credible but incorrect things. All right, so that's all 10 predictions. I'm not sure we're up to with scoring. I think we missed two. Is it three? Joe? Yeah. I think you were keeping count better of the scoring.

Six or seven?

Seven.

I think we think we're on seven. Maybe seven and a half if we give you half. Yes. It was better than random anyway, so. All right, wonderful. So just to wrap up, that was the. That was the review of last year's predictions, and next episode, we'll go on to making new predictions for 2026. All right, see you again both shortly.

Richie, thank you.

Thank you, Richie. This was fun.