DeepMind CEO Demis Hassabis suggests that AI is overhyped in some respects
2 min read
DeepMind’s Demis Hassabis compares the current frenzy in AI investment to the craze in crypto, expressing concern about “grifting.”
DeepMind’s Demis Hassabis is experiencing a mix of excitement and apprehension due to the current AI funding frenzy. As the CEO of DeepMind, a leading AI research company, Hassabis understands the transformative potential of artificial intelligence. However, he also acknowledges the hype cycle that AI has entered.
In an interview with the Financial Times, Hassabis noted, “In a way, AI’s not hyped enough but in some senses it’s too hyped…We’re talking about all sorts of things that are just not real.”
The influx of investor funding into generative AI startups reminds Hassabis of other hyped-up areas, such as cryptocurrency. He finds this trend unfortunate, stating, “And it clouds the science and the research, which is phenomenal.”
His concerns about grifting are supported by the numbers. PitchBook data shows that almost $30 billion has been invested in 691 generative AI deals in 2023. Investors are eager to capitalize on the AI trend and are investing heavily.
Hassabis’ warnings are justified. In March, the SEC settled charges against advisors who were accused of falsely claiming to use AI. With such large sums of money involved, some individuals will try to exploit the hype cycle.
Despite his concerns, Hassabis is highly optimistic about AI’s future impact: “I think we’re only scratching the surface of what I believe is going to be possible over the next decade-plus. We’re at the beginning, maybe, of a new golden era of scientific discovery.”
He highlights DeepMind’s AlphaFold project as evidence. Released in 2021, it has predicted over 200 million protein structures, significantly speeding up biological research globally. According to Hassabis, this demonstrates how AI can accelerate scientific research.
Founded in 2010, DeepMind’s goal has been to achieve artificial general intelligence (AGI) that matches human cognition. While some experts suggest that AGI may still be decades away, Hassabis believes there’s “about a 50 percent chance” it will be achieved within the next 10 years. He notes that this timeline hasn’t shifted much since DeepMind’s inception.
