Hassabis highlights how social media algorithms amplified misinformation and polarization, urging AI creators to avoid repeating these mistakes and to prioritize societal benefits.

Recent advances, such as the DolphinGemma model for decoding animal communication, showcase AI's positive potential, but Hassabis emphasizes the need for responsible innovation.

Hassabis advocates for international cooperation and regulatory standards to oversee AI's integration into critical sectors like healthcare and science, balancing innovation with risk mitigation.

DeepMind's projects like the Scalable Instructable Multiworld Agent (SIMA) aim to create adaptable AI, but reliance on natural language instructions raises concerns about misuse in real-world scenarios.

He stresses that AI should serve and support people rather than manipulate attention, advocating for responsible innovation and regulation to ensure positive societal impact.

He estimates AGI could emerge within 5-10 years but warns society is unprepared, calling for global collaboration to manage risks such as job displacement and ethical dilemmas.

He calls for a thorough, scientific approach to AI development, prioritizing testing, understanding implications, and human well-being over profits to ensure responsible deployment.

A summit involving DeepMind and OpenAI highlighted economic risks like inequality, with Hassabis advocating for AI to address grand societal challenges while avoiding social media's profit-driven pitfalls.

Google DeepMind CEO Demis Hassabis warns that AI development must learn from social media's harmful patterns, emphasizing responsibility, transparency, and ethical considerations to prevent societal harm.

Hassabis discusses AI's current limitations, noting its proficiency in complex tasks like math Olympiads but struggles with basic problems, a phenomenon called 'jagged intelligence' that hampers true artificial general intelligence (AGI) development.

Hassabis stresses the importance of balancing opportunities and risks in AI, especially as it approaches general intelligence, with a focus on understanding and mitigating potential harms.

He warns that AI models designed to maximize attention could worsen issues like addiction and mental health crises if not properly regulated, echoing concerns about social media's impact.