Italy’s Minerva-3B, trained from scratch on 2.5 trillion tokens, scored just 4.9% on Italian school exams, raising questions about native-language investment.
Minerva. The opposite path.
Browsing Category
Artificial Intelligence
305 posts
The Forecast Is the Plan.
Major AI labs and investors are publicly committing to automating AI R&D, with specific targets and plans, signaling a shift in industry strategy.
AMÁLIA · The Three Hard Questions.
Portugal’s €5.5M AMÁLIA model is operational, but key structural questions remain about openness, native data, and goals, impacting future AI policy.
The Regulatory Vacuum.
Google discloses a zero-day exploited by criminals, exposing gaps in U.S. AI regulation and security frameworks amid political and policy uncertainties.
China Sphere Capability Gap, Q2 2026 Update: Five Labs, Five Strategies, One Narrowing Frontier
Chinese labs launched five frontier-tier models in April 2026, narrowing the capability gap with the US but maintaining economic and strategic advantages.
The Compounding Error Problem — Why 99.9% Alignment Decays to 60% in 500 Generations
Analysis of how 99.9% alignment accuracy drops significantly over multiple AI generations, raising concerns about recursive self-improvement safety.
The Co-Founder’s Black Hole — A Structural Read on Jack Clark’s Automated AI R&D Essay
Jack Clark predicts over 60% chance of fully autonomous AI research by 2028, raising concerns about institutional capacity and future risks.
Every Benchmark Launched 2023-2024 Has Fallen — The METR / SWE-Bench / CORE-Bench / MLE-Bench / PostTrainBench Sequence
Every major AI research benchmark launched between 2023-2024 has either saturated or is nearing saturation, indicating rapid progress in AI capabilities.
Jack Clark Says It Out Loud — Reading the Co-Founder’s 60%/2028 Estimate on Automated AI R&D
Anthropic’s co-founder Jack Clark publicly estimates a 60% chance of autonomous AI R&D by 2028, signaling significant policy implications.
The Continual Learning Research Map: Where the Memento Constraint Stands in May 2026
Six months after initial analysis, the research on the Memento Constraint confirms it remains a key bottleneck, with no fully ready solutions yet. Deployment expected by 2028-2030.
