AI Science: Google I/O Unveils New Possibilities

At its recent annual developer conference, 'I/O,' Google opened new horizons in technology by presenting novel possibilities for scientific research leveraging artificial intelligence (AI). During the event's keynote, Demis Hassabis, CEO of Google DeepMind, stated that AI stands at the cusp of the 'singularity' – the theoretical future point where AI surpasses human intelligence – heightening expectations for the present and future of AI science.

Google has continued to develop AI tools for specific scientific problem-solving, while also demonstrating their utility by announcing achievements in applying AI to address real-world societal issues like climate change prediction. Pushmeet Kohli explained that AI is advancing beyond merely accelerating science to independently conducting scientific research, envisioning a future where LLM-based agent systems can perform advanced research without human intervention. Over the past year, AlphaFold's protein structure predictions have been utilized by over 3 million researchers worldwide.

Amidst the popularity of these AI tools, signs of recalibration in enthusiasm and resource allocation are also emerging. Google's subsidiary Isomorphic Labs secured $2 billion in Series B funding to leverage AlphaFold and related technologies for drug discovery. However, The Los Angeles Times reported last month that John Jumper, a Google Fellow who received a Nobel Prize for AlphaFold, is now involved in AI coding tasks rather than science-specific AI tools. Google's reputation has recently suffered due to performance issues with its coding tools, making the deployment of top talent to solve these coding challenges a logical step.