Shane Legg on AGI, the 2028 Turning Point, and Future of Human Intelligence

Shane Legg's AGI Prediction

• 💡 Shane Legg, co-founder of Google DeepMind and coiner of "AGI," predicts a critical turning point in intelligence.
• 🎯 He estimates a 50% probability of human-level AGI by 2028, defining AGI as a machine capable of all typical human cognitive tasks.
• 🧠 Intelligence is viewed as a spectrum, not binary, with current AI excelling in language but still developing in visual reasoning and continuous learning.
• ✅ AGI will emerge when AI no longer makes human-like mistakes, leading to the birth of a digital intellect equal to humanity.

The Hardware Gap: Human Brain vs. AI

• ⚡ The human brain's neural signals travel at only 30 meters per second, while AI processes information at near light speed (300,000 km/s).
• 🔌 Our brain operates on a mere 20 watts, whereas modern data centers consume over 200 megawatts and weigh millions of kilograms.
• 📈 AI processes data at 10 GHz, significantly faster than the human brain's 100-200 Hz frequency.
• 🔬 Legg highlights that machines are 6-8 orders of magnitude ahead in key intelligence dimensions like energy, space, bandwidth, and signal speed.

Economic and Social Restructuring

• ⚠️ The advent of AGI will cause a structural collapse of the economy, making complex projects feasible with significantly fewer human professionals.
• 💼 Intellectual labor, especially remote-friendly professions like education, law, and medicine, will be fundamentally reshaped by AI.
• 💰 The traditional model of labor-for-resources may become obsolete as AI becomes the primary producer, leading to a "democratization of intelligence."
• 🌍 This shift will shake thousands of years of human social structure, necessitating new societal contracts.

Preparing for the AI Era

• 🚀 Despite the challenges, Legg remains a technological optimist, believing AI can solve intractable problems and bring unprecedented prosperity.
• 🔑 He emphasizes the urgency for humanity to act now and redesign its future, warning against complacency, especially among experts.
• 🛡️ "System 2 Safety"—designing AI to monitor its own thought processes and align with human values—is crucial for controlling superintelligence.
• 🌱 We are the first generation to live with machines that surpass human intelligence, requiring proactive preparation for this new era.