Scientists and researchers at Google DeepMind, the research arm of Alphabet Inc., are building new artificial intelligence (AI) systems that will advance science, transform work and improve the lives of billions of people, two senior researchers at the lab told The Business Standard in an interview in Bengaluru.
“Google DeepMind's mission is to build AI responsibly to benefit humanity,” said Sesh Ajjalap, senior director at Google DeepMind.
Click here to connect with us on WhatsApp
Ajjalap, who is based in the US, was recently in India to participate in the Google I/O Connect Bengaluru 2024 event. He said, “What you are seeing today is short-term (innovation). But we are taking a longer-term view and tackling some of the toughest and most quintessential problems of our time.”
Google DeepMind brings together two leading AI labs, Google Brain and DeepMind, into one dedicated team that has delivered some of the biggest research breakthroughs over the past decade, many of which are foundational to today's AI industry.
One of the innovations Ajalap highlighted was AlphaFold, an AI system that accurately predicts 3D models of protein structures, sparking a new wave in biology. Google DeepMind recently announced the third major version of AlphaFold, which helps scientists design drugs and precisely target diseases.
Researchers have long searched for clean, unlimited sources of energy, and nuclear fusion, the same process that powers the stars in the universe, is one solution.
Breaking down hydrogen, a component common in ocean water, and fusing it together releases vast amounts of energy. On Earth, scientists have been replicating the extreme conditions of stars using tokamaks, doughnut-shaped devices surrounded by magnetic coils that contain hydrogen plasma hotter than the core of the sun. But the plasma inside a tokamak is unstable, making sustaining the process needed for fusion a tall order.
To solve this problem, DeepMind worked with EPFL's Swiss Plasma Centre to develop the world's first deep reinforcement learning (RL) system that autonomously predicts how to control the coils and contain the plasma, opening up new avenues in nuclear fusion research.
“With a mobile-first population and a burgeoning startup and developer ecosystem, India is uniquely positioned to drive the next generation of AI innovation,” Ajarapu said, citing innovations by Indian companies. From consumer experience to agriculture to social enterprises, AI is tackling some of the biggest challenges in sectors and industries.
Manish Gupta, director at Google DeepMind, said inclusive AI is a key theme for the company's India team. “We're developing AI in a way that benefits billions of people, especially those who are not impacted by it,” he said. “A lot of our work is about linguistic inclusivity and multicultural awareness, making sure all these models understand different cultural nuances.”
The Google DeepMind India team spoke about its work on language solutions to help computing developers working on AI. This includes expanding Project Vaani, a collaboration with the Indian Institute of Science (IISc), to provide developers with more than 14,000 hours of speech data in 58 languages. Gupta said the project is in the middle of Phase 2 and is expanding to cover 160 districts across Indian states.
DeepMind tracks possible biases in its computing models as part of its commitment to responsible AI. The company has made some of its data sets and benchmarks publicly available so other researchers can study the biases. “And of course, we're devising mitigation techniques in our language models to gain a richer cultural understanding so we don't perpetuate these biases,” Gupta said.
