No, AI systems cannot feel self-aware or conscious in any way, says Google DeepMind CEO Demis Hassabis

Google DeepMind co-founder and CEO, Demis Hassabis , has said that AI systems cannot feel self-aware or conscious in any way. In a recent interview, he warned that as AI becomes more autonomous, keeping these systems safe, aligned, and firmly under human control is essential for long-term success. Hassabis argues that instilling ethics in AI must be approached with the same care and intentionality as raising a child. Though technical progress continues to grow, current AI models still lack any genuine subjective awareness or self-experience. With stakes rising alongside AI’s growing capabilities, the Nobel Prize awardee emphasises that ethical alignment and governance must keep pace with innovation.


What Google DeepMind CEO Demis Hassabis said about AI consciousness

In a recent interview with CBS’s Scott Pelley, Hassabis noted that his DeepMind team’s work isn’t explicitly aimed at producing conscious or self-aware AI.

“I don’t think any of today’s systems feel self-aware or conscious in any way,” he said. However, he clarified that he still doesn’t rule out the possibility that “these systems might acquire some feeling of self-awareness.”

“That is possible,” he added.

Hassabis claimed that what matters even more is that systems grasp the concepts of “self” and “other”—an initial step toward more advanced cognitive abilities. He also highlighted the importance of guiding these systems’ development so they remain consistent with human values.

"Teaching morality to AI—much the way you would teach your child. Can we make sure that they are doing what we want, that they benefit society, and that they stay on guardrails—with safety limits built into the system?” Hassabis explained.

He mentioned that DeepMind’s AI models are built with architectural goals but acquire capabilities through data-driven learning.

“We don’t program that in. It learns like a human being would learn,” Hassabis continued. This approach can yield unexpected behaviours, prompting initiatives like Project Astra to study how these systems adapt. Hassabis even highlighted the rapid pace at which AI development is progressing.

“It's moving incredibly fast. We’re on some kind of exponential curve of improvement,” he said while predicting that Artificial General Intelligence may emerge in five to ten years, with deeply nuanced systems integrated into daily life by 2030.

“We’ll have a system that really understands everything around you in very nuanced and deep ways. I think it will have a breakthrough moment in the next couple of years,” he said, foreseeing breakthroughs in robotics in the coming days.

As AI’s impact grows, Hassabis also emphasised the need for international cooperation on regulation, ethics, and accountability. He argued the real challenge for this decade is to keeping these systems “safe, useful, and aligned with the values of the people they serve.”