In a recent interview with the Financial Times, Google DeepMind CEO Demis Hassabis said Google is feeling confident about its standing in the artificial intelligence race. However, he made it clear there is still intense competition.
Hassabis noted that Google’s latest model, Gemini 3, was a major upgrade, pointing to the model’s positive user feedback.
Instead of getting distracted by the competition, Hassabis claimed that Google’s priority is to constantly improve its research and ensure it is reflected in its products.
“It’s a ferocious, intense competition,” Hassabis said, explaining that their strategy is to simply “block out the noise and just execute.”
A major highlight during the interview was multimodal AI, which Hassabis described as one of Google’s key strengths. Unlike traditional models that focus on text, multimodal systems can work with images, video and audio simultaneously.
In Hassabis’ view, this is key to building assistants that can function in real-world settings.
“That’s what you would need for a kind of assistant that travels around with you in the real world,” he said, mentioning phones, smart glasses and even robotics as areas where this technology could have a major impact.
Hassabis framed DeepMind as the “engine room” of Google, responsible for building advanced models and then finding ways to integrate them into everyday products.
The interview also touched on concerns about AI safety and misuse. “Society is right to be worried about these things,” Hassabis said, stressing that companies with large user bases have a responsibility to act carefully.
He also mentioned that one method to reduce backlash is by clearly showing how AI can be used for good, especially in areas like science and medicine.
Hassabis denoted AlphaFold and drug discovery research as examples where the benefits of AI are easier to see and harder to argue against.
In regards to money and hype in the industry, Hassabis took a balanced view.
He admitted that some parts of the AI market do look overheated, particularly when it comes to startups with no products to show.
At the same time, he said that despite the demand for AI tools growing rapidly, companies are still struggling to keep up with demand due to limited resources.
When asked about the possibility of an AI bubble, Hassabis said it is not a simple yes-or-no question. He said that even if some investments fail, the technology itself is not going anywhere and that he does not agree with claims that AI is unrealistic.
He explained that human intelligence is already general and believes AGI is still possible, though likely five to 10 years from now.
“I think 2030 is probably the earliest,” he said while cautioning that breakthroughs often take longer than expected.
Hassabis said Google is focused on long-term research and slowly building AI into real products. He said the goal is to create systems that work, are built responsibly and are useful to people.
