AI regulation and hallucinations: Lessons from Generative AI London

This article brings together highlights from one of our exclusive events, you can catch all the best moments from our summits OnDemand right here.

The Generative AI Summit in London brought together some of the brightest minds in the industry to explore the challenges, breakthroughs, and potential of this game-changing technology. One standout moment was the panel discussion featuring Tom Mason, Chief Technology Officer at Unlikely AI.

We caught up with Tom after his session to dig deeper into the insights from his panel and hear more about the exciting work Unlikely AI is doing. Here are the key highlights from our conversation. Check out the full interview here:

Just looking for a quick read? We compiled the key highlights from Tom’s expert panel below.

1. The complex world of AI regulation

AI regulation is a hot topic, and Tom’s panel tackled it head-on. He shed light on how different regions—like the US, EU, UK, and China—are approaching regulation and what that means for AI innovation.

• Finding common ground: Tom emphasized the importance of building trust between AI creators and users. Regulations that strike the right balance can create a solid foundation for scalable and responsible AI adoption.
• Innovation vs. red tape: While strict regulation can hold back innovation, a lack of oversight risks eroding trust. Tom’s takeaway? It’s all about finding that sweet spot.

Every region has its own approach, and while it’s still early days, the discussions happening now are setting the stage for the future of AI.

2. Solving the hallucination problem in generative AI

One of the biggest challenges for generative AI today is managing “hallucination”—when AI confidently generates inaccurate or outright fabricated information. Tom shared how Unlikely AI is tackling this head-on with a unique approach.

• A new kind of architecture: Unlikely AI combines symbolic world models with large language models (LLMs) to create what Tom calls a “compound system.” This setup allows for greater control over hallucination, so the AI can switch between creative and hyper-accurate modes depending on the use case.
• Grounded in facts: By rooting their models in trusted data sources like Wikipedia, they’re working to ensure accuracy without sacrificing the flexibility generative AI is known for.

Tom explained that this issue is a major barrier to scaling AI solutions in high-trust environments like enterprise applications, but Unlikely AI is on a mission to fix it.

3. Why London is an AI hotspot

As a London-based company, Unlikely AI is putting the city on the generative AI map. Tom highlighted what makes London (and the UK) such a great place for AI innovation.

• Pro-innovation vibes: The UK is in a unique position—still light on regulation but heavy on support for experimentation, making it an ideal environment for cutting-edge AI development.
• Talent galore: With a deep pool of diverse expertise not just in London but across the UK, it’s the perfect breeding ground for cross-functional, collaborative teams.

It’s clear that London isn’t just keeping up with global AI hubs—it’s carving out a space at the forefront.

4. What’s next for Unlikely AI?

Unlikely AI is gearing up for some big moves, and Tom gave us a sneak peek at what’s coming in the next year:

• Early-stage validation: In Q1 2025, the team plans to roll out initial Proof-of-Concepts (PoCs) with companies in different sectors to test their technology in high-throughput environments.
• Full launch: By mid-2025, Unlikely AI is set to bring its solutions to market, offering businesses a way to deploy generative AI with accuracy and confidence.

If your business is looking to leverage the next generation of generative AI, Tom is more than open to having a conversation.

Wrapping things up, Tom spoke passionately about the role of community in driving AI forward. Events like the Generative AI Summit bring together builders, engineers, and executives to share ideas, build connections, and foster innovation.

“It’s all about bringing different skill sets together,” Tom said. “Communities like AI AI are doing an incredible job at creating the spaces we need to collaborate and push the industry forward.”