For all their remarkable fluency with language, code, and complex reasoning, the latest generation of AI systems — including GPT-5 — have stumbled on one of psychology's most enduring measures of basic cognition: the sustained attention test. Researchers found not a gradual decline but a near-total collapse in performance as cognitive demands increased, revealing that the ability to hold focus over time may rest on principles fundamentally different from those powering today's most capable models. The finding invites a quieter, more unsettling question beneath the headlines about AI progress:
Google News
· Jun 23, 2026
· technology
· 1 min de leitura
High
Apuração: Verificado
Advanced AI Models Collapse on Classic Psychology Test, Raising Questions About Human-Level AI
◉
Technology
8777 artigos relacionados
Nvidia Q3 earnings loom as options traders brace for 7.69% volatility …
The Economic Times · 100% correspondência
Apple's Siri AI Leverages Google Gemini While Plotting Deep App Integr…
digitimes · 99% correspondência
Trump's Q1 2026 Stock Trades Reveal $220M-$750M in Big Tech Bets
News18 · 99% correspondência
Nvidia Shield Users Report Dark Video, CEC Glitches, Overheating Issue…
bgr.com · 99% correspondência
Google Pixel 11 Pro Fold launches with Tensor G6, Gemini AI, and new O…
Business Today · 99% correspondência
Cobertura Relacionada
Deutsche Welle · Aug 25
Australia's music charts ban fully AI-generated tracks, allow AI-assisted songs
Australia's ARIA charts will exclude fully AI-generated music starting August 31, while permitting AI-assisted recording…
News-Medical · Aug 25 Study identifies most reliable epigenetic clocks for testing longevity interventionsA meta-analysis of 51 longevity studies identifies which epigenetic aging biomarkers most reliably respond to interventi…
Sahi · Aug 25 Kaynes Tech Positioned for Growth as India's ₹62,500 Cr Mobile Scheme LaunchesKaynes Technology sees 40% revenue growth amid India's ₹62,500 crore Mobile Phone Manufacturing Scheme, while global chi…
The Irish Times · Aug 25 Irish-led study to unlock sleep-mental health connectionIrish-led DataGenChrono programme will investigate genetic and biological pathways connecting sleep disruption to depres…
Viés e Enquadramento
Não há dados de análise detalhada para esta lente. Tente executar as lentes novamente no painel de administração.
Impacto Geopolítico
🌎
AI capability gaps in sustained attention tasks have minimal geopolitical impact; primarily a technical limitation affecting AI development timelines rather than international power dynamics.
Escalation Risk
low
Confidence
HIGH
No direct shifts. Indirectly relevant: countries investing heavily in AI (US, China, EU) may experience delayed timelines for AGI-dependent strategic advantages, potentially extending current technological leadership windows.
Lente Econômica
📈
AI model limitations on attention tasks may delay human-level AI development, potentially affecting AI-dependent sectors and investment timelines in the technology industry.
Market Signal
mixed
Confidence
MED
Technology/Software
Artificial Intelligence
Venture Capital/Investment
Enterprise Software
Automation Services
Consumers may experience delayed deployment of fully autonomous AI systems in critical applications (healthcare, autonomous vehicles, financial services), potentially extending reliance on human oversight and hybrid AI-human solutions longer than anticipated.
Regulators may use these findings to justify more cautious AI deployment frameworks and extended testing requirements before approving AI systems for high-stakes applications. Could support arguments for stronger AI safety regulations and slower rollout timelines.