Sir Demis Hassabis learned this week that he had won the Nobel Prize in Chemistry when his wife – also a scientific researcher – received several Skype calls urgently asking for his phone number.
“My head was completely exhausted, which rarely happens. It was. . . almost like an out-of-body experience,” said Hassabis, co-founder and chief executive of Google DeepMind, the substitute intelligence division of the Silicon Valley search giant.
The Nobel Prize in Chemistry, which Hassabis shared together with his colleague John Jumper and US biochemist David Baker, was won for solving an inconceivable problem in biology that had remained unsolved for 50 years: predicting the structure of each protein known to mankind using a well known AI software as AlphaFold.
Having overcome this long-standing challenge with far-reaching implications for science and medicine, Hassabis has set his sights on climate change and healthcare. “I would like us to assist solve some diseases,” he told the Financial Times.
His team is working with drugmakers Eli Lilly and Novartis on six drug development programs focused on disease areas resembling cancer and Alzheimer's. Hassabis said he expects to have a drug candidate in clinical trials inside two years.
His other major focuses are using AI to more accurately model the climate and pushing the last word frontier of AI research: inventing machine intelligence on par with human intelligence.
“When we glance back in 10 years, I hope that (AI) has ushered in a brand new golden era of scientific discovery in all of those different areas,” said Hassabis, a former neuroscientist and video game designer. “That’s what got me into AI in the primary place. I see it as the last word tool for accelerating scientific research.”
The DeepMind duo were recognized on Wednesday, a day after former Google colleague and veteran AI scientist Geoffrey Hinton won the physics prize alongside physicist John Hopfield for his or her work on neural networks, the foundational technology for contemporary AI systems, that support healthcare, social media and the self – driving – and AlphaFold itself.
The recognition of AI breakthroughs marks a brand new era in research and underscores the importance of computational tools and data science in solving complex scientific problems on far shorter timescales, from physics to mathematics to chemistry and biology.
“It's obviously interesting that the (Nobel) committee decided to make such a press release by bringing the 2 together,” Hassabis said.
The awards also reflect the guarantees and potential pitfalls of AI.
Hopfield and Hinton were pioneers of the discipline within the early Eighties. Hinton, who’s 76 and left Google last yr, said he had no plans to analyze further. Instead, he desires to advocate for work on the safety of AI systems and be certain that governments support them.
In contrast, the DeepMind pair won for work presented primarily within the last five years and stays extremely optimistic about its societal impact.
“The impact of (AI) on science specifically, but in addition on the trendy world on the whole, is now very, very clear,” said Maneesh Sahani, director of the Gatsby Unit at University College London, a research institute that focuses on machine learning and theory focuses on neuroscience. Hinton was founding director of Gatsby's in 1998, while Hassabis worked there as a postdoctoral fellow in 2009 and eventually spun DeepMind out of the UCL institute in 2010.
“Machine learning is popping up in all places, from people analyzing ancient texts in forgotten languages ​​to X-rays and other medical imaging techniques. We now have a toolbox that may advance science and academic disciplines in all possible directions,” said Sahani, who can be a neuroscience professor.
The latest iterations of AlphaFold have “implications across medicine, biology and lots of other fields” because they’re so fundamental to living organisms, said Charlotte Deane, a professor of structural bioinformatics on the University of Oxford.
“Many were skeptical once they began, but in a short time their program outperformed all other protein structure prediction programs,” said Venki Ramakrishnan, a biologist who won the Nobel Prize in Chemistry in 2009 for his work on protein synthesis. “It really modified the sphere dramatically.”
AlphaFold has been utilized by greater than 2 million scientists to, amongst other things, analyze the malaria parasite, develop a vaccine, improve plant resistance to climate change, and study the structure of the nuclear pore – one in all the biggest protein complexes within the human body.
Rosalyn Moran, a neuroscience professor at King's College London and managing director of AI start-up Stanhope AI, said: “Tool making is scientific work by employees.” . They are sometimes the unsung heroes of science. For me, that was probably the most exciting a part of the award.”
AlphaFold still has flaws, its developers reported earlier this yr, including “hallucinations” of “false structural order” in cell regions which might be actually disordered. Another challenge in using AI for scientific research is that some necessary areas of research could also be less extensive than protein evaluation on experimental data.
In their work on the Physics Nobel, Hinton and Hopfield used fundamental concepts from physics and neuroscience to develop AI tools that may process patterns in large information networks.
The Boltzmann machine invented by Hinton could learn from concrete examples somewhat than instructions. The machine was then capable of recognize recent examples of categories it had been trained on, resembling images of cats.
This variety of learning software, often known as neural networks, now forms the premise of most AI applications, resembling facial recognition software and huge language models, the technology that underlies ChatGPT and Google's Gemini. One of Hinton's former students, Ilya Sutskever, was co-founder and chief scientist of ChatGPT maker OpenAI.
“I’d say I'm someone who doesn't really know what field he's in, but would really like to know how the brain works,” Hinton, a pc scientist and cognitive psychologist, said during a news conference this week. “And in my attempts to know how the brain works, I even have helped develop technology that works surprisingly well.”
The AI ​​awards have also dropped at the fore the interconnectedness of scientific discoveries and the necessity to share data and expertise – an increasingly rare phenomenon in AI research that happens inside business entities resembling OpenAI and Google.
Neuroscientific and physical principles were used to develop today's AI models, while data generated by biologists contributed to the invention of AlphaFold software.
“Scientists like me have traditionally solved protein shapes using complex experimental methods that may take years,” said Rivka Isaacson, professor of molecular biophysics at King's College London and one in all AlphaFold's first beta testers. “However, it was these solved structures that the experimental world provides for public use that were used to coach AlphaFold.”
She added that AI technology has allowed scientists like her to “dive deeper into the function and dynamics of proteins, ask different questions, and potentially open up entirely recent areas of research.”
Ultimately, AI – like electron microscopy or X-ray crystallography – stays an analytical tool and never an independent agent conducting original research. Hassabis insists that technology cannot replace the work of scientists.
“Human ingenuity comes into play – asking questions, conjectures, hypotheses, our systems can’t do any of that,” he said. “(AI) is currently just analyzing data.”
