Demis Hassabis: Why AGI is Bigger than the Industrial Revolution & Where Are The Bottlenecks in AI

By 20VC

Categories: VC, Startup

Summary

Demis Hassabis estimates AGI arrival within 5 years with 'very good chance'—quantifying it as 10x the Industrial Revolution at 10x the speed. Compute remains the primary bottleneck, not just for scaling but for testing new algorithmic ideas, while scaling laws show diminishing but 'still substantial' returns.

Key Takeaways

  1. AGI definition: a system exhibiting all cognitive capabilities of the human mind. The brain is the only existence proof of general intelligence, making it the crucial benchmark for measuring progress toward AGI.
  2. Deep Mind predicted ~20 years from 2010 to AGI through compute and algorithmic progress extrapolation. Current trajectory suggests they're 'pretty much on track,' implying AGI feasibility within 5 years from this 2024 interview.
  3. Compute bottleneck has dual impact: scaling systems (larger architectures) AND enabling experimentation. Researchers need substantial compute to test algorithmic ideas at reasonable scale before mainline deployment.
  4. Scaling laws aren't plateauing—returns remain 'very substantial' though less exponential than early gains. Frontier labs continue getting strong returns from compute expansion despite slower performance doubling per generation.
  5. Critical missing capability: continual learning. Current systems don't learn post-training in the real world, representing a major gap versus human-like AGI. This architectural limitation hasn't been solved because 'people haven't quit' attempting it.

Topics

Transcript Excerpt

I would say about 90% of the breakthroughs that underpin the modern AI industry were done either by Google brain or Google research or deep mind. So one of our groups the returns are kind of still very substantial although they're a bit less than they were obviously at the start of all of this scaling. We have amazing guests on the show but very few honestly will be considered in the same realm as Newton Turing Einstein. Our guest today is one of the greatest minds on the planet and I consider m...