AI & other lessons from the Nobels
ONE technology that is linked to both the physics and chemistry Nobel prizes this year is artificial intelligence (AI). The physics laureates — John Hopfield and Geoffrey Hinton — applied principles and tools of physics to develop methods that have facilitated machine learning. Hopfield used the principle of atomic spin to create a structure that can store and reconstruct information, while Hinton invented a method that can independently discover properties in data forming the basis for the large artificial neural networks now in use. These discoveries, over the decades, have helped powerful computers mimic human functions like memory and learning.
There is no substitute for basic research. This point has to be re-emphasised in the Indian context because we bemoan that no Indian has got a Nobel for scientific research done in India after CV Raman.
The chemistry Nobel has gone to David Baker of the University of Washington; Demis Hassabis and John M Jumper of Google DeepMind. Baker has been awarded for his work on computational protein design and the DeepMind researchers get the other half of the prize for developments in protein structure prediction. Proteins are essential for cellular functions in living bodies. They are intricately folded, depending on their atomic structure and water molecules surrounding them. In a single protein, there could be trillions of potential interactions, creating countless possible shapes. Baker developed computational tools for designing new proteins, which, in turn, has opened the door for new therapies and treatments. The Google DeepMind researchers have developed an AI system called AlphaFold that predicts the three-dimensional structure of proteins from their amino acid sequences.
The physics, chemistry and medicine Nobel prizes again demonstrate that fundamental research is vital for knowledge creation and there is no substitute for basic research. This point has to be re-emphasised in the Indian context because every time, we bemoan that no Indian has got a Nobel for scientific research done in India after CV Raman. Hargobind Khorana, S Chandrasekhar and Venki Ramakrishnan won their Nobels for research conducted in foreign universities. Along with applied research and technology development, sustained investments are needed in fundamental research for breakthroughs.
The Nobel is usually given for pathbreaking work that leads to transformative tools and technologies. Hopfield, who is 91, began working on what came to be known as the 'Hopfield network' and associative memory in the 1980s. So did Hinton, with his 'Boltzmann machine' method. Their methods enabled the machine learning revolution decades later, in 2010. And finally, we have consumer products like ChatGPT. Similarly, chemistry laureate Baker has been working on protein structures for decades. He came up with the first tool for protein structure prediction, Rosetta, in 1998.
The edifice of AI is based on many building blocks created over the decades. Not many know that Indian scientists have also contributed to this process. At a time when AI was just emerging as a possibility and engaging the attention of scientists along with the development of digital computers in the 1950s, physicist-turned-statistician Prasanta Chandra Mahalanobis developed a concept which came to be known as the Mahalanobis Distance. It helps detect outliers in data and measure dissimilarities in data points. Subsequently, Mahalanobis Distance found wide applications in computer science and AI. At the Indian Statistical Institute (ISI) founded by him, Mahalanobis recognised the importance of cybernetics and invited its pioneer Norbert Wiener to spend time in the institute as a visiting professor in 1955. Wiener initiated ISI researchers like Dwijesh Datta Majumder to work in pattern recognition, fuzzy logic and neural networks.
In 1966, Raj Reddy, a young Indian doctoral student in America, pioneered the development of systems for recognising continuous speech. He developed a new system called Hearsay I for continuous speech recognition. Along with giving computers the capability to memorise and learn like humans, making them capable of recognising human speech was a critical step towards achieving AI. Reddy later developed Hearsay II, Harpy and Dragon systems, which are the basis of commercial speech recognition technology used in computers as well as robots.
Reddy's fundamental discovery was the 'blackboard model' for coordinating multiple knowledge sources and it has found wide applications in AI, like voice control of robots and speaker-independent speech recognition. For his contributions, Reddy was given the Turing Award — considered the Nobel of computer science — in 1994. His contribution to the building blocks of AI is as seminal as that of 2024 Nobel winners.
The Nobel for AI-related discoveries comes with warnings about the potential danger of the technology from the two pioneers. Hinton has called AI chatbots 'quite scary' after he quit the position he held in Google's AI division. He also fears that the widespread use of AI could widen economic disparities in society as the increased productivity and wealth due to AI would only help the rich. AI could gobble up a lot of mundane jobs.
To redress such impacts of AI on inequality, Hinton has said that the governments should consider a system of universal basic income for those impacted. Similarly, Hopfield has signed petitions calling for strong controls of technology and risk and benefit analysis of AI. He fears new applications of AI could lead to a dystopian society imagined by George Orwell.
Coming back to India's Nobel drought, we need to introspect. The Central government has finally operationalised a new mechanism to fund research, the Anusandhan National Research Foundation, but the funding pathways are still unclear. If India aspires to make a mark in the world of science, basic research in universities and labs should get adequate and continuous support. This is a long-term game. The focus on applied research, incremental innovation and technology development helps meet immediate societal and industrial needs.
So, we will have to strike a balance between basic research and technology while allocating resources. The private sector, too, should draw lessons from this year's Nobel. The chemistry Nobel has been shared by researchers from Google, reflecting the deep investments the company has made in basic research. Whichever way one looks, there is no magic wand or recipe for getting a Nobel other than sustained investments in basic research.
