While Still Lagging Behind OpenAI, There Is One Area Where Google’s AI Has Taken the Lead: The Number of Nobel Prizes

Google has just achieved a significant milestone the company should be proud of. On Wednesday, Demis Hassabis and John Jumper, who work in Google’s DeepMind division, were awarded the Nobel Prize in Chemistry for their work on protein structure prediction. Their use of the AI model AlphaFold to make these predictions has been a standout achievement in a year where AI has been a dominant force. The Nobel Prize isn’t only a triumph for DeepMind but also a significant win for Google.

Google 1 - OpenAI 0. The recent recognition puts Google ahead in its competition with OpenAI, given that the tech giant now has more Nobel Prizes to its name. What’s more, the award from the Royal Swedish Academy of Sciences strongly validates Google’s efforts in the AI department and shows confidence in the future of AlphaFold.

A notable acquisition. Google purchased DeepMind in January 2014 and has been providing ongoing support for the division, which has primarily focused on academic and research pursuits rather than practical applications. Google’s achievements have been remarkable, starting with AlphaGo, followed by AlphaZero, and most recently, AlphaFold, which has reached its third major release.

Two sides of the same coin. Google’s backing has been crucial for DeepMind to continue advancing its work. Concurrently, the company is also striving to lead the AI race in the business and consumer sectors, but that’s a completely different story.

A respite for Gemini. It’s interesting to see that while DeepMind products have a great reputation, Google’s end-user AI still seems to struggle. This was evident during the launch of Bard and its AI image-generator. However, the rollout of Gemini, also developed by DeepMind, appears to have stabilized after some initial problems and shows great promise.

Two isn’t better than one. For several years, Google Brain and DeepMind operated separately. However, in April 2023, Alphabet merged them into a single division focused on developing AI solutions. It’s still too early to determine whether this merger has had a positive impact, given that the lines of work appeared to be independent until now.

Could Transformer be a Nobel Prize winner of the future? In 2017, researchers from Google Brain and Google Research (notably, none from DeepMind) published a well-known paper on Transformer. The “new simple network architecture” sparked the generative AI revolution and led to the development of major AI models such as OpenAI, DALL-E (for images), and ChatGPT (for text).

The economic and social impact of these models has been more prominent than that of AlphaFold. However, they still make mistakes and have hallucinations, which affects their scientific credibility. Their ongoing development shows a promising improvement, and as some experts suggested, this technology might be considered for the Nobel Prize in the distant future.

Another Nobel laureate from Google (sort of). On Tuesday, Geoffrey Hinton was named one of the Nobel Prize in Physics winners. Hinton is widely regarded as one of the pioneers of AI, particularly in the field of deep learning and neural networks. Interestingly, Hinton worked at Google’s AI division for many years before leaving to focus on his recent work, which involves warning about the potential risks of AI. In fact, Hassabis has also expressed reservations about AI.

Image | Focal Foto | Greg Bulla