The State of AI in 2025 – AI Expert Andrew Grill

Andrew delivered the opening keynote at the London AI Beach Party titled: The State of AI in 2025

He then engaged with the audience answering 8 very curious questions about AI over a 21-minute period following his 40-minute keynote.

1. “Do a lot of AI rely on OpenAI’s ChatGPT sort of in the background, because my understanding is a lot of them run off of one OpenAI is the original. If OpenAI went down today, what would happen to majority of the other AI, because they’re not all coded independently, is what I think I don’t know. So say that went down, what would happen to Bob, for example, or Google’s VEO3, would they what would happen?”
2. “I’m curious about the authentication of what’s real and what’s not in the future. And I wonder how much blockchain or other technologies will play into that?”
3. “If trust is the currency of the digital age, what might be the interest rate? So what will grow that trust or erode it?”
4. “How do you know which version to ask, given that the answer is deferred quite a lot, is it kind of short of trial and error? How do you save time and use it effectively?”
5. “What about the sustainability side of all of this? Because we get a lot of information about how OpenAI Sam said, you know, just saying please and thank you cost a lot in terms of the energy usage. So the balance between curiosity, want to play, want to experiment, want to learn, versus how many forests am I decimating with me playing around with making a picture of action AI me, or, you know, studio Ghibli me, or, you know, all of those things. So how do we balance that?”
6. “Is it possible that you can manipulate, for example, the generative AIs like ChatGPT, by posting a big amount of fake information on the internet?”
7. “An AI and hiring report from Insight Global found that 62% of UK hiring managers say AI tools are rejecting more qualified candidates. Are we prioritising system efficiency over attracting diverse talent, those who may think differently but more innovatively, and what steps are we now taking to improve bias within AI?”
8. “If someone is, for example, trying to positively discriminate because somebody’s used AI in, say, a job application or something like that. Other than, sometimes it gives you kind of like, overly obsequious, okay, if someone’s trying to work out whether someone has used AI to positively discriminate if they have, then, other than it, sometimes phrasing it so, like, overly obsequiously. How can you actually positively, if you want to do that, how can you see that someone’s used AI in, say, a job application?”