Nobel Prizes highlight the intersection of AI and fundamental research



Most of us tend to see AI in purely ‘tech’ terms. But this year’s Nobel Prizes in physics and chemistry underline the importance of the ‘science’ behind the tech. They have also brought to the fore the role of AI in basic research, while also muddying the disciplinary boundaries among the branches of scientific inquiry. John Hopfield used principles of physics to mimic the human brain in artificial neural networks. Geoffrey Hinton used statistical modelling and psychology to recreate how associative memories are made because connections are reinforced when pieces of a network work together. The pair, considered among the godfathers of AI, won the physics prize. Then, in chemistry, Demis Hassabis and John Jumper won the Nobel for having utilised AI to predict the structure of almost all known proteins. The physics prize, in effect, went to work on computing and mathematics, while the chemistry prize was derived through computing.

Hopfield is a physicist whose area of work has been microbiology. Hinton has worked in psychology and quit Google to be able to air his concern over AI. Hassabis has been a chess player and a computer programmer before turning his attention to biochemistry. AI is, thus, blurring the distinctions among areas of scientific inquiry where computing and data are figuring out solutions to problems where maths is beyond human capacity. This could lead to a combination of two outcomes in fundamental research. One would be to use AI to make incremental hops in our understanding of nature. The other would be to ask even more difficult questions and then deploy the computing technology that helps arrive at an answer.

The caveat here is AI has to be trained properly to be able to deliver the research outcomes we are seeking. At each stage of problem-solving, the programming needs will mount. AI, as it is now, is not a very good expert at everything. It hallucinates, incapable of telling truth from falsehood, an imprecise tool at best. Humans will need their eureka moments for quite some time longer.



Source link