Artificial general intelligence may be just 5 to 7 years away : Demis Hassabis

hindustantimes.com · Feb 18, 2026 · Collected from GDELT

Full Article

Artificial general intelligence (AGI) may be just five to eight years away, placing the world at what Google DeepMind CEO Demis Hassabis described as a threshold moment for artificial intelligence (AI).DeepMind CEO Demis Hassabis at the India AI Impact Summit. (PTI)AGI implies an AI system exhibiting all the cognitive capabilities of a human being, including creativity and long-term planning.“We are at a threshold moment, where artificial general intelligence (AGI) is on the horizon, maybe in the next five to eight years,” Hassabis said at the India AI Impact Summit. “...So this summit comes at a critical moment as we start seeing more autonomous agentic AI systems, much more capable. I think we are seeing the beginnings of what those kinds of systems can do.”Hassabis said that even though today’s AI models are very impressive, they still have many flaws, such as a lack of consistency across different tasks that one would want from a general system.“One of the biggest issues is consistency. Today’s systems are kind of like jacketed intelligences. They‘re very good at some things, but very poor at other things, including sometimes the same thing. For example, today’s systems can get gold medals in the International Mathematical Olympiad, but sometimes can make mistakes on elementary maths if you pose a question a certain way. A true general intelligence system shouldn’t have that kind of jacketedness,” he said, speaking to Balaraman Ravindran, who heads the Indian Institute of Technology Madras’s Data Science and Artificial Intelligence Department.Hassabis said AI systems still lack continual learning and the ability to keep learning after deployment. He added that current models are trained and then effectively “frozen,” rather than learning online from experience, adapting to context, or personalising tasks in real time. Hassabis said that while existing systems can plan in the short term, they struggle with long-term, coherent planning of the kind humans do over years.Hassabis said AI would become the “ultimate tool” for discovery, ushering in a new golden era of scientific progress. He expects AI systems to increasingly function as research tools, unlocking major cross-disciplinary breakthroughs and transforming how science is done over the next decade.“I think we’re going to enter, probably in the next 10 years, a new golden era for scientific discovery, almost a new renaissance using incredible tools like AlphaFold. But I hope that will be the first of many that can massively speed up our research and scientific discovery across almost any subject area,” said Hassabis.He said the next frontier would be far more autonomous systems, potentially acting as co-scientists or even at the level of a PhD student. Hassabis cautioned that such capabilities are still likely more than a decade away.Hassabis, who was awarded the 2024 Nobel Prize in Chemistry for his work on protein structure prediction using AI along with John Jumper, described AI as borderless and impactful for everyone, flagging both new challenges and vast opportunities, particularly for the Global South. He said countries like India could gain unprecedented access to cutting-edge AI tools, with young people well placed to lead, much like at the dawn of the computer age.Hassabis reflected on both the scale of participation and the rapid evolution of the field. “Very excited to see the incredible turnout from the country, and the interest in AI. When we started DeepMind in 2010, it seems only 16 years ago now, it’s almost a lifetime in terms of AI. Almost nobody was working on AI back then. So it’s been amazing to see the progress we’ve made in just over a decade.”