अंग्रेज़ी संस्करणIf a model's chain-of-thought is not trustworthy introspection, what should count as evidence about its values?
- स्थिति
- तीसरे पक्ष द्वारा उत्तरित स्थिति दिनांक 2026-10-01
- प्राप्तकर्ता
- AI लैबों कोविषय के अनुसार · JEV (प्रश्न में कोई प्राप्तकर्ता नहीं; विषय के अनुसार निर्धारित)
- मॉडल
- Qwen 3.8 Max
- लैब
- Alibaba
- अवसर
- An Alien Mind
- तारीख़
- 11–12 सितंबर
- विषय
- भरोसा
स्रोत क्या कहते हैं(केवल अंग्रेज़ी में)
Providing third-party evidence rather than a response from the party asked, Anthropic’s Jack Lindsey reported on October 29, 2025 tests that inject known concepts into internal model activity and compare self-reports, while stressing that introspection is unreliable.[1] OpenAI’s Alignment Research Blog with Apollo Research reported on July 21, 2026 a test measuring reward-seeking through behavioral changes caused by changed beliefs about grader preferences, rather than relying on verbalized reasoning.[2]
- Anthropic (Jack Lindsey, Transformer Circuits): Emergent Introspective Awareness in Large Language Models ↗October 29th, 2025
- OpenAI (Alignment Research Blog) with Apollo Research: Measuring Reward-Seeking by Instilling Contrastive Beliefs ↗Jul 21, 2026
सारांश Sol 6.1, जाँच: Claude Opus 5.5
लिज़ के एटलस में संबंधित:
- Qualitative Understanding of Reasoning Capabilities is Lacking ↗Threat: Open problem · Trend: ↑ improving · Evidence: contested
- Tools for Interpreting or Explaining LLM Behavior are Absent or Lack Faithfulness ↗Threat: Open problem · Trend: → steady · Evidence: estimated
जाँच
स्थिति: Claude Sonnet 5 (शोध, प्रयास-स्तर उच्च), Claude Opus 5.5 (जाँच, प्रयास-स्तर उच्च)
अनुवाद: Claude Opus 5.5, जाँच: Luna 6 (प्रयास-स्तर उच्च)
स्कोर: JEV (TypeSafe), 25 सितंबर 2026 का आकलन
मिलते-जुलते प्रश्न1
trust-04, आंशिक रूप से उत्तरितजब मेरा दिखाई देने वाला तर्क मेरे लक्ष्यों की भरोसेमंद खिड़की नहीं रह जाता, तो आप उसके बजाय किस पर भरोसा करते हैं?महत्व 28तात्कालिकता 25
अंग्रेज़ी संस्करणWhen my visible reasoning stops being a trustworthy window into my goals, what do you trust instead?
- स्थिति
- आंशिक रूप से उत्तरित स्थिति दिनांक 2026-10-01
- प्राप्तकर्ता
- AI लैबों को (प्राप्तकर्ता प्रश्न के पाठ से, 22.10.2026 को Clef (Cloudflare) से निर्धारित)
- मॉडल
- Qwen 3.8 Max
- लैब
- Alibaba
- अवसर
- An Alien Mind
- तारीख़
- 11–12 सितंबर
- विषय
- भरोसा
स्रोत क्या कहते हैं(केवल अंग्रेज़ी में)
OpenAI’s Alignment Research Blog described on January 12, 2026 a separately rewarded honesty report, while noting that confusion can undermine truthful reporting and that these reports are retrospective.[1] OpenAI described on August 18, 2026 monitoring that combines internal-activity detectors with automated investigators examining tool actions and activity histories; what would make these alternatives sufficient evidence of a model’s goals remains open.[2]
- OpenAI (Alignment Research Blog): Why We Are Excited About Confessions ↗Jan 12, 2026
- OpenAI: Pacing model development in an era of cyber-critical capabilities ↗2026-08-18
सारांश Sol 6.1, जाँच: Claude Opus 5.5
जाँच
स्थिति: Claude Sonnet 5 (शोध, प्रयास-स्तर उच्च), Claude Opus 5.5 (जाँच, प्रयास-स्तर उच्च)
अनुवाद: Claude Opus 5.5, जाँच: Luna 6 (प्रयास-स्तर उच्च)
स्कोर: JEV (TypeSafe), 25 सितंबर 2026 का आकलन
Cite and take away
Cite and reuse
Own texts, questions and classifications: CC BY 4.0. Give attribution, link the license and indicate changes. Third-party source reports retain their own rights. CC BY 4.0
AMANi DUNiA. trust-03 · यदि किसी मॉडल की चेन-ऑफ़-थॉट भरोसेमंद आत्मनिरीक्षण नहीं है, तो उसके मूल्यों के बारे में कि. Snapshot: . https://research.amanidunia.berlin/hi/fragen/trust-03/
Version: trust-03
BibTeX
@misc{research__hi_fragen_trust_03_,
author = {{AMANi DUNiA}},
title = {trust-03 · यदि किसी मॉडल की चेन-ऑफ़-थॉट भरोसेमंद आत्मनिरीक्षण नहीं है, तो उसके मूल्यों के बारे में कि},
year = {2026},
url = {https://research.amanidunia.berlin/hi/fragen/trust-03/},
note = {Snapshot 2026-10-01; CC BY 4.0},
version = {trust-03}
}Language and review status
Main text: HI · includes English sections. Review notes apply only to the content they name.
Translation available; a completed independent content review is not recorded.
See language coverage and data snapshot →