Shift in AGI Projections
François Chollet, the creator of the Abstraction and Reasoning Corpus (ARC-AGI) and a prominent figure in artificial intelligence research, has revised his timeline for the arrival of Artificial General Intelligence (AGI). Historically recognized as a skeptic regarding the immediate capabilities of Large Language Models (LLMs) and rapid scaling, Chollet’s updated outlook suggests that AGI may arrive sooner than his previous estimates.
According to reports circulating on social media platforms such as Reddit, Chollet’s projections have undergone several significant shifts. In 2024, Chollet estimated a timeline of approximately ten years for the realization of AGI. By February 2026, he had adjusted this estimate to roughly 2030, aligning the milestone with the anticipated development of ARC-AGI versions 6 or 7.
Impact of Recent Technological Developments
The most recent adjustment to Chollet’s timeline follows the release of the Astra model and the reported saturation of ARC-AGI-3 benchmarks. When questioned whether his 2030 estimate remains accurate in light of these developments, Chollet indicated that AGI could arrive “sooner, given progress is happening faster than I expected.”
While these comments reflect a notable shift in sentiment from a researcher known for cautious reasoning, it should be noted that these statements originated from discussions on Reddit and have not been independently verified through formal academic publications or official corporate press releases.
Benchmarking and Industry Context
The saturation of ARC-AGI-3 benchmarks suggests a shift in how AI capabilities are being measured, moving toward more complex reasoning tasks. The ARC-AGI framework has long been used to test an AI’s ability to solve novel problems that require reasoning rather than mere pattern recognition.
The acceleration in development timelines has prompted various speculative discussions within the tech community. Some observers have questioned whether the current influx of capital into AI development is driven by the potential for AI to solve complex biological challenges, such as human aging, despite broader economic or environmental uncertainties. However, these remains speculative observations from online commentators and lack empirical confirmation.