Demis Hassabis says society is “not ready” for the AI future

AI development must prioritize human-centric scientific advancement, particularly in medicine, where tools like AlphaFold demonstrate the potential to solve complex biological challenges. While recursive self-improvement offers significant acceleration, closed-loop systems without human oversight present substantial risks, necessitating rigorous engineering, third-party auditing, and international safety standards to prevent catastrophic incidents. Intelligence and consciousness remain distinct phenomena; therefore, current efforts should focus on building controllable, intelligent tools rather than attempting to create conscious entities. This approach allows for a safer, more deliberate progression toward AGI. Demis Hassabis, co-founder of Google DeepMind, emphasizes that global collaboration and robust, hardened sandboxes are essential to navigate the competitive race for AI supremacy while maintaining safety and alignment with human values.