The AI Futures Project presents a compelling scenario analysis forecasting the development of artificial general intelligence (AGI) and superintelligence by 2027. The report, authored by forecasting experts including Daniel Kokotajlo and Eli Lifland, maps out a plausible path from today's AI assistants to autonomous agents that could fundamentally transform society within the next few years.
Key Highlights
- Mid-2025: First generation AI agents emerge as personal assistants, capable of tasks like ordering food and managing spreadsheets, though still unreliable
- Late 2025: Companies like 'OpenBrain' build unprecedented datacenters with 2.5M GPU-equivalents, training models with 1000x more compute than GPT-4
- AI development focuses on creating systems that can accelerate AI research itself, potentially leading to rapid recursive improvement
- The authors predict superintelligence could arrive by decade's end, with impacts exceeding the Industrial Revolution