DeepMind CEO Demis Hassabis has made a significant prediction about artificial general intelligence (AGI), suggesting that by 2025, AI models will demonstrate ‘cognitive consistency’ - a crucial stepping stone toward AGI. In his assessment, cognitive consistency refers to AI systems maintaining coherent reasoning and behavior across different tasks and contexts, similar to human intelligence. Hassabis emphasizes that current AI models, despite their impressive capabilities in specific domains, lack this fundamental attribute of general intelligence. The CEO’s timeline aligns with Google’s broader AI development strategy and their commitment to responsible AI advancement. He notes that achieving cognitive consistency would represent a major milestone in AI development, though it wouldn’t necessarily mean full AGI has been achieved. The prediction comes amid increasing competition in AI development between major tech companies and growing discussions about AI safety and regulation. Hassabis also addresses the importance of careful development and testing protocols as AI systems become more sophisticated. The article highlights how DeepMind’s approach combines neuroscience insights with machine learning techniques, aiming to create more robust and generalizable AI systems. While some experts view this timeline as optimistic, Hassabis’s position as leader of one of the world’s premier AI research organizations lends weight to his prediction.
Source: https://www.businessinsider.com/google-deepmind-ceo-demis-hassabis-agi-consistency-2025-8