A good friend shared yesterday the interview video of Sergey attending the All-in Summit 2024. Boss Sergey is very optimistic about the future of AI, whether it's about how AI will change various fields or Google's advantages in the AI competition, his confidence is very firm.
Video link: https://www.youtube.com/watch?v=XzK9bx3CSPE
Content:
(0:00) David Friedberg intros Sergey Brin (1:41) What Sergey is working on at Google (5:45) Is Google chasing a "God Model"? (8:49) Thoughts on the massive AI chip buildout (12:54) Future of human-AI interaction (14:58) Changing Google's conservative product culture (18:21) Is the "Race to AGI" overblown?
Main text
David Friedberg, when introducing Sergey Brin, called him one of the greatest entrepreneurs of our times. He not only has an innovative mindset but is also willing to take on things that seem impossible.
Sergey Brin has gradually stepped back from the day-to-day leadership at Google, but now he is re-engaging in Google's artificial intelligence field to support its AI strategy. Brin said that he feels very fortunate to be engaged in work that he finds meaningful, helping people access information.
AI is advancing at an incredible speed
Brin said that he almost works at Google every day, focusing on the development of AI technology, so he was somewhat "unwilling" to come out for interviews. Brin marveled at the rapid development of AI technology, especially in the past few years, where the pace of progress in AI has been incredible. Compared to when he was a graduate student in the 90s, AI was just a minor part of computer science courses. Today, AI has undoubtedly become the core of the tech field. There are new developments every month, and as a computer scientist, one doesn't want to miss out.
The impact of AI on search engines
Brin stated that while AI will affect the way searches are conducted, its scope of application goes far beyond the realm of search, such as in program writing and image generation. Writing code from scratch is much more difficult compared to letting AI do it. Now even Mr. Brin himself directs AI to write code and uses AI to generate and solve Sudoku puzzles. He also lightly criticized Google engineers for not using AI enough.
The Future of AI Models
Brin believes that while there are currently multiple specialized AI models handling different tasks, the future trend may shift towards a more unified model architecture. He is uncertain whether a so-called "God Model" will emerge, but he trusts that the integration of AI models is the direction for the future.
Computational Demand
Brin agrees with the surge in computational demand, and Google has been rapidly expanding its computing capabilities to meet the needs of both internal use and cloud clients for AI computation. Although he has reservations about external expectations regarding the speed of AI development, he firmly believes in the enormous potential of AI applications across various fields, with limitless possibilities for the future.
The Impact of Artificial Intelligence in Different Fields
Especially the perspective on biology and robotics. He mentioned that Google's AlphaFold has become an important tool in the field of biology since its launch, and almost every biologist is using it. Although this is a different type of AI, all AI technologies seem to be gradually converging.
In terms of robotics, Brin said we are still in the "wow" stage. Although some general language models have demonstrated the ability to make robots complete tasks, they have not yet reached sufficient robustness to become a commonly used tool in daily life. However, he believes this field is advancing rapidly.
Google acquired and sold several robotics companies in its early days, including Boston Dynamics. Although these technologies looked very cool, the AI technology at the time was not enough to support their continued development. Nowadays, with the rise of multi-modal AI models that can understand vision, images, and scenes, new possibilities for the future of robotics have emerged.
The future way of human-computer interaction
Brin currently finds it difficult to accurately predict developments five years from now. He used a term, there is some "watercooler chitchat" (gossip between colleagues at work), because the technical foundation of AI is evolving rapidly. Many applications are developed from some simple demonstrations, and these demonstrations often lead to amazing innovations. However, moving from demonstration to large-scale production applications still requires a great deal of effort and time.
Google's Cultural Shift
Regarding how to be bolder in launching AI products. Although Google may be somewhat conservative internally, especially considering that language models might make mistakes in their early stages, Brin advocates for being willing to take certain risks and tolerate some embarrassment 😓, because AI technology has enormous potential and magic.
Competition in the AI Field
Brin believes that despite fierce competition among AI companies, it is beneficial for the industry's development. He mentioned that Google has ranked high in recent analytical tasks (He does care lol😄), and he takes pride in Google's progress in the AI field. Despite the intense competition, he believes that AI technology will bring great value and transformation to humanity, just as the internet once changed the world.