A prominent AI researcher, Daniel Kokotajlo, who previously worked at OpenAI, has revised his earlier prediction about when artificial intelligence might reach autonomous capabilities that could pose existential risks. He had previously suggested AI could independently write code and begin self‑improving as early as 2027, potentially leading to superintelligence with severe unintended consequences. He now believes such milestones are likely further off, shifting the timeline into the early 2030s.
This updated view effectively moves the projected arrival of transformative AI toward around 2034. Kokotajlo has not provided a new estimate for when AI might pose an actual threat to humanity, signaling a move away from near‑term apocalyptic framing. His revised perspective reflects a recognition that the path to advanced autonomous systems is more complex and uncertain than earlier speculative models suggested.
Experts note that while AI capabilities continue to expand rapidly, integrating these systems into complex real-world environments introduces unpredictable delays and limitations. True artificial general intelligence remains a debated concept, and many researchers question whether current AI architectures can evolve into self-directed superintelligent agents without major breakthroughs in both hardware and theory.
Despite the delayed timeline, AI development continues at a fast pace, with leading labs and companies pursuing increasingly capable systems. Some experts emphasize the need for robust safety research, governance, and regulatory frameworks, while others highlight AI’s economic and societal benefits. The discussion underscores the tension between optimism for technological progress and caution over potential long-term risks.
