Position : La fenêtre d'action préventive avant l'arrivée d'IA surhumaines sur la plupart des tâches cognitives se referme rapidement
Year: 2026
Type: preprint
Abstract: In this position paper, we argue that without coordinated intervention, AI systems will by default exceed human expert performance on most measurable cognitive tasks by 2030. Using a joint hierarchical Bayesian model, we forecast frontier performance trajectories across 60 benchmarks spanning reasoning, mathematics, coding, scientific knowledge, and agentic capabilities. We find that nearly all current benchmarks (98% in our set) are on track to saturate within four years. Security-critical benchmarks (cybersecurity, autonomous AI R&D, biology, chemistry) show even faster trajectories, saturating before 2028. While these findings may not constitute definitive proof of imminent AGI on their own, they add to a converging body of evidence indicating that superhuman performance on cognitive tasks is approaching faster than commonly assumed. Acknowledging that we neither understand nor control the behavior of current GPAI models, this timeline leaves limited runway before irreversible impacts. We outline implications for global governance and international coordination, AI safety research, and evaluation practices.
Source: HAL
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7139335562