· organisation
Wikipedia (opens in new tab)

Demis Hassabis on OpenAI

Acknowledging competitor (strong)

TL;DR

Demis Hassabis credits OpenAI and others for successfully scaling AI technology that originated with Google research.

Key Points

  • He credited OpenAI and others for successfully scaling the AI chatbots built upon the 2017 transformer technology developed by Google researchers.

  • Hassabis argued that claims of current AI systems possessing “PhD intelligences” are "nonsense" due to missing reasoning capabilities.

  • Google DeepMind has recently adopted a more entrepreneurial, faster methodology to compete with rivals in the current 'ferocious competitive environment.'

Summary

Demis Hassabis, CEO of Google DeepMind, acknowledges that while Google researchers developed the foundational transformer technology in 2017, the company was comparatively slow to commercialize and scale the innovation. He specifically credits OpenAI and other competitors for effectively using this technology to bring their AI chatbots to a broad user base. Hassabis stated that the core challenge for Google was execution rather than initial invention, noting that competitors did very well in the commercialization and scaling phases of the technology.

This perspective surfaces amid intense competition in the AI sector, forcing DeepMind to adopt a scrappier, faster approach, acting as the primary engine for Google's AI product launches. Furthermore, Hassabis has been vocal in pushing back against industry claims, particularly those from OpenAI, regarding current AI systems achieving genuine "PhD-level" intelligence, asserting that critical reasoning capabilities are still fundamentally missing from today's models.

Key Quotes

“There were only a few dozen researchers working on AI properly and we all knew each other. We were, in most cases, really good friends,”

Frequently Asked Questions

Demis Hassabis views OpenAI as a major competitor that effectively commercialized foundational AI technology originating from Google research. He credits them for scaling chatbots successfully, which Google was slower to do.

His core stance has been one of competitive acknowledgement; he recognizes their success in productizing AI while maintaining a critical view on claims of AGI parity. There is no indication of a fundamental shift, but rather a continuous competitive appraisal.

Demis Hassabis publicly dismissed claims that OpenAI's models, like GPT-5, possess true "PhD-level" intelligence. He argues that these systems lack the general reasoning capabilities of top human scientists.