· concept
Wikipedia (opens in new tab)

Demis Hassabis on Quantum Computing

AI will outpace quantum (strong)

TL;DR

Demis Hassabis conjectures that advanced AI on classical systems may solve problems thought to require quantum computers, while keeping an eye on quantum progress.

Key Points

  • He conjectures that classical AI systems can model complex systems, potentially performing tasks anticipated for quantum computers, as evidenced by AlphaFold's success with proteins.

  • He keeps a close eye on quantum computing advances, sometimes having bets with quantum friends about whether AI or quantum will solve a problem first.

  • The belief is that AI, running on classical Turing machines, can discover and model patterns in nature more efficiently than previously thought possible.

Summary

Demis Hassabis holds the position that classical computing systems, empowered by advanced Artificial Intelligence, possess the potential to efficiently model and solve complex systems, including those previously thought to be the exclusive domain of quantum computers. He bases this conjecture on the success of tools like AlphaFold in approximating solutions for quantum-level phenomena, such as protein folding, suggesting that the utility of quantum computing might be reduced by continued AI advancement. The Chief Executive Officer and Co-founder of Google DeepMind maintains an active interest in the field, viewing it as a potential avenue for scientific acceleration.

While presenting this view that AI could make some quantum computing applications obsolete, he acknowledges the ongoing advancements in the field, even maintaining friendly bets with colleagues regarding whether AI or quantum computing will achieve certain milestones first. He notes that problems thought to require quantum simulation, like protein folding, were successfully approximated by classical machine learning models. However, he continues to closely monitor quantum computing systems, recognizing their power and acknowledging that AI may eventually be used to perfect quantum systems, rather than completely replace them.

Frequently Asked Questions

Demis Hassabis believes that the continuing advancement of Artificial Intelligence on classical systems might allow for the efficient modeling of problems currently thought to require quantum computers. He has publicly stated his conjecture that AI could make some quantum computing applications obsolete before they fully materialize.

He has suggested that AI might be able to solve problems for which quantum computers were previously thought necessary, citing the success of AlphaFold on a quantum-scale problem like protein folding. However, he maintains an interest in quantum progress and suggests AI may ultimately help perfect quantum systems, rather than entirely replace them.

Demis Hassabis sees quantum computing as an interesting area of progress, but he maintains an ongoing 'bet' that AI on classical machines will reach certain computational milestones first. He believes that current AI systems can already model quantum systems, reducing the perceived necessity for quantum hardware in certain scientific domains.