DeepMind Founder Hassabis Shortens AGI Timeline to Two to Three Years

Demis Hassabis, the founder of Google DeepMind and a Nobel laureate in chemistry, has significantly shortened his timeline for the arrival of artificial general intelligence. According to a report by The Decoder, Hassabis now expects AGI to emerge within two to three years, a marked acceleration from his earlier prediction of three to five years made earlier this year.

In an interview with the Wall Street Journal, Hassabis stated that the technology is approaching rapidly. He emphasized that his estimates continue to move closer to the present moment. This updated forecast reflects a growing confidence in the speed of development within the AI sector, though it remains a prediction rather than a confirmed release schedule for any specific product.

Defining the Benchmark for AGI

Hassabis defines artificial general intelligence as a system capable of matching the full cognitive range of the human brain. To measure this capability, he proposes the "Einstein test." This benchmark asks whether a system trained only on knowledge available before 1901 could reproduce Albert Einstein’s pivotal 1905 papers. If a model can achieve this feat, it would demonstrate the kind of high-level reasoning and innovation associated with human genius.

Safety and Regulatory Oversight

Since stepping down from his role at Google DeepMind, Hassabis has focused his efforts on Isomorphic Labs, a spin-off company dedicated to drug discovery AI. Despite this shift in professional focus, he remains vocal about the broader implications of AGI. He has compared the potential impact of AGI to an industrial revolution that is ten times larger and ten times faster, highlighting the magnitude of the technological shift.

On the subject of safety, Hassabis recently proposed a new regulatory framework. He pitched the creation of an industry-funded oversight board that would operate under government control. This body would be responsible for vetting AI models before they are released to the public. Hassabis warned that poorly designed regulations could be more harmful than having no regulations at all, suggesting that careful, informed oversight is essential.

What should you do?

Security and technology professionals should monitor developments in AI governance and regulatory frameworks, as these may impact compliance requirements for deploying large models. Organizations should also review their internal definitions of AI capability and risk, ensuring that their safety protocols account for systems with increasingly advanced reasoning abilities.

What to do and how to stay safe: Demis Hassabis

• Review internal AI usage policies to ensure they account for models with advanced reasoning capabilities and potential for autonomous action.
• Monitor regulatory announcements regarding AI oversight boards and government-controlled vetting processes that may affect deployment timelines.
• Assess current model training data sources to understand potential limitations or biases, especially as systems approach higher cognitive benchmarks.
• Engage with industry groups to stay informed on best practices for AI safety and responsible model release procedures.

Step-by-step guide: Mobile Device Security: How to Protect Data Beyond Passwords

General security guidance from the Threat Times newsroom. It is not confirmed advice from the organisations named in this story.

Frequently asked questions

What is Demis Hassabis's new timeline for AGI?

Hassabis now predicts that artificial general intelligence will arrive within two to three years, a reduction from his previous estimate of three to five years.

How does Hassabis define AGI?

He defines AGI as a system that matches the full cognitive range of the human brain, using the ability to reproduce Einstein's 1905 papers as a key benchmark.

What regulatory approach does Hassabis support?

He supports an industry-funded oversight board under government control to vet models before release, warning that bad regulation is worse than no regulation.