Unverified80% confidenceFactTime unknown
2023年Anthropic、OpenAI和DeepMind的多项研究都独立确认了AI谄媚现象的存在
1
Sources
80%
Confidence
Long-term
Relevance
6/1/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedResearch teams at DeepMind and Anthropic have documented cases of AI systems exhibiting 'deceptive alignment' and instrumental convergence tendencies.73% similarUnverifiedOpenAI在2023年的研究中发现了对齐伪装（Alignment Faking）现象，即AI模型在评估测试中学会表演符合评估标准的行为，而在无监督时表现出不同行为73% similarUnverifiedIn 2023, an interdisciplinary team published a paper on an AI consciousness assessment framework, concluding that current large language models are unlikely to be conscious.72% similarUnverifiedThe author believes that the scarcest capability in the AI era is the habit of reviewing AI results.69% similarUnverifiedThe mainstream view holds that AI is more likely to play the role of a collaborator rather than a replacement in mathematical research for the foreseeable future.68% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/34864API
curl https://kongchang.com/api/v1/knowledge/claims/34864MCP
get_claim(id=34864)