Leading AI expert delays timeline for its possible destruction of humanity

2026-01-06 · Source: AI (artificial intelligence) | The Guardian · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Emerging Technologies & Innovation, AI Ethics & Safety · Depth: Novice, short

Summary

Daniel Kokotajlo, a prominent AI expert and former OpenAI employee, has revised his timeline for AI achieving autonomous coding and subsequent superintelligence, pushing back his initial 2027 prediction. Kokotajlo's "AI 2027" scenario, released in April, posited that unchecked AI development could lead to a superintelligence that outfoxes world leaders and destroys humanity by mid-2030. This scenario sparked significant debate, drawing both admirers and detractors, including references from US Vice-President JD Vance and criticism from NYU professor Gary Marcus. The updated forecast now places autonomous coding in the early 2030s and superintelligence around 2034, omitting a prediction for human destruction. This revision reflects growing doubts among experts about the imminence of Artificial General Intelligence (AGI) and the practical complexities of real-world integration, despite goals from companies like OpenAI to develop automated AI researchers.

Key takeaway

For AI scientists and research teams assessing future AI capabilities, you should re-evaluate aggressive timelines for autonomous coding and superintelligence. The revised forecast from Daniel Kokotajlo, pushing autonomous coding to the early 2030s and superintelligence to 2034, suggests that practical integration challenges and the "jagged" nature of AI performance introduce significant delays. Factor these real-world complexities into your strategic planning and risk assessments, rather than relying solely on theoretical "intelligence explosion" models.

Key insights

An AI expert has delayed his timeline for AI achieving autonomous coding and superintelligence, citing real-world complexities.

Principles

• AI performance is often jagged, not linear.
• Real-world inertia delays societal change from AI.
• Integrating advanced AI faces complex strategic challenges.

In practice

• Consider "AI 2027" as a scenario for discussion.
• Evaluate AI timelines with real-world integration challenges.
• Monitor OpenAI's internal goals for automated AI researchers.

Topics

• AI Timelines
• AI Safety
• Autonomous Coding
• Superintelligence
• Artificial General Intelligence

Best for: AI Scientist, Research Scientist, AI Ethicist, Policy Maker, AI Researcher

Related on AIssential

• What the Freakiness of 2025 in AI Tells Us About 2026 — AI progress in 2025 highlighted both advanced reasoning and generative capabilities, alongside challenges in trust and general intelligence.
• Experts give a 50% chance of AGI by 2050, 75% chance by the mid-2060s, and 95% chance by 2090 — Experts predict AGI has a 50% chance by 2050, rising to 95% by 2090.
• AIs can now often do massive easy-to-verify SWE tasks and I've updated towards shorter timelines — AI progress on easy-to-verify tasks is accelerating superexponentially, shortening timelines for AI R&D automation.
• It is the process of rapidly ever improving differentiation between noise and signal patterns and constant generalization of those that produces intelligence, not merely compression of data. [D] — True intelligence stems from continuous pattern differentiation and generalization, not just data compression.
• The next 36 months will be WILD — Expert consensus points to AGI/ASI by 2027-2028, driven by super-exponential AI progress and an irreversible industrial race.
• Google's SHOCKING "POST AGI" paper... — Google DeepMind projects Artificial Superintelligence (ASI) as a likely outcome within 1-2 decades, viewing AGI as a mere starting point.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by AI (artificial intelligence) | The Guardian.