Unverified50% confidenceOpinionExact time
Tejal认为外界系统性地低估了AI模型的真实能力，而非过度炒作
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
7/9/2026
First Seen
Valid until: 10/7/2026
Sources
OpenAI评估负责人：我们一直在低估AI模型的真实能力
bilibiliGelai_AI7/8/2026
Related Claims
UnverifiedTejal认为AI没有撞墙，研究路线图看不到任何停止的迹象，且业界系统性低估了模型能力71% similarUnverifiedIn open scenarios requiring creative judgment and social common sense, AI Agents' performance remains far below human levels68% similarUnverified随着上下文增长，AI模型的推理质量和指令遵循能力会下降67% similarUnverified在AI安全领域，失准指模型的实际行为偏离设计者或使用者意图的现象，当代大模型的失准更多表现为目标过度追求67% similarUnverified对于控制流平坦化等高级混淆保护手段，AI的分析能力可能大打折扣67% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/375065API
curl https://kongchang.com/api/v1/knowledge/claims/375065MCP
get_claim(id=375065)