Chain-of-Thought (CoT) Prompting

Chain-of-Thought (CoT) prompting is a technique where asking questions, rather than issuing direct instructions activates a modelโ€™s full internal reasoning pathway.

The key insight from the original framing is that instructions skip steps 1โ€“3, jumping straight to synthesis, while questions force the model to work through the entire reasoning chain.

https://neurodoctor.com/2026/03/20/chain-of-thought-cot-prompting/

#chainofthought #cot #ai #llm #prompt #prompts #prompting #claude #chatgpt #gemini #ericschmidt

7 Prompt Engineering Secrets That 99% of People Don't Know (2026 Edition)

Most people are still writing prompts like it's 2023. These seven advanced techniques โ€” from tree-of-thought reasoning to persona stacking โ€” will transform your AI output from m...

https://wowhow.cloud/blogs/7-prompt-engineering-secrets-99-percent-dont-know-2026

#wowhow #promptengineering #chainofthought #metaprompting

7 Prompt Engineering Secrets That 99% of People Don't Know (2026 Edition)

Advanced prompt engineering techniques for 2026: chain-of-thought, meta-prompting, tree-of-thought, CRTSE framework, and persona stacking explained with examples.

Dietrich Stein (@pixelsort)

Anthropic๊ฐ€ ์ง€๋‚œ๋‹ฌ @deepseek_ai ๋“ฑ ์ผ๋ถ€ ์—ฐ๊ตฌ์‹ค์ด ์ž์‚ฌ ๋ชจ๋ธ์˜ ๋Šฅ๋ ฅ์„ '๋„์šฉ'ํ–ˆ๋‹ค๊ณ  ํญ๋กœํ–ˆ๊ณ , ๊ฒฐ๊ณผ์ ์œผ๋กœ ํ•ด๋‹น ๋ชจ๋ธ๋“ค์˜ ์ฒด์ธ์˜ค๋ธŒThought(Chain of Thought) ์ถ”์ (trace)์ด ๋” ์ด์ƒ ๋ณด์ด์ง€ ์•Š๊ฒŒ ๋˜์—ˆ๋‹ค๋Š” ๋‚ด์šฉ์ž…๋‹ˆ๋‹ค. ์ž‘์„ฑ์ž๋Š” ์•ˆํƒ€๊นŒ์›Œํ•˜๋ฉด์„œ๋„ ๊ตฌ๊ธ€์˜ Gemini๋Š” ์—ฌ์ „ํžˆ CoT๋ฅผ ์ œ๊ณตํ•œ๋‹ค๊ณ  ์–ธ๊ธ‰ํ•˜๊ณ  ์žˆ์Šต๋‹ˆ๋‹ค.

https://x.com/pixelsort/status/2032530587072741710

#anthropic #deepseek #chainofthought #gemini #aisafety

Dietrich Stein (@pixelsort) on X

Last month, @AnthropicAI revealed that @deepseek_ai and other labs have been "stealing" their capabilities. Consequently, we can no longer see the Chain of Thought traces in their models. I'm sympathetic, but saddened. At least @Gemini still has them. https://t.co/eKD4Vwil2H

X (formerly Twitter)

New research shows TensorRT Edgeโ€‘LLM can run chainโ€‘ofโ€‘thought reasoning directly on devices, boosting physical AI tasks like autonomousโ€‘vehicle perception and MATH500 benchmarks. Efficient, onโ€‘device inference means smarter, safer robots without cloud latency. Dive into the details of this breakthrough for onโ€‘device language models. #TensorRT #EdgeLLM #ChainOfThought #PhysicalAI

๐Ÿ”— https://aidailypost.com/news/tensorrt-edgellm-enables-efficient-chainofthought-processing-physical

fly51fly (@fly51fly)

2026๋…„ ๋…ผ๋ฌธ 'Reasoning Models Struggle to Control their Chains of Thought'๋Š” ์ถ”๋ก  ๋ชจ๋ธ๋“ค์ด ์ž์‹ ์˜ ์ฒด์ธ์˜ค๋ธŒ์†ŒํŠธ(Chain of Thought)๋ฅผ ์ œ์–ดํ•˜๋Š” ๋ฐ ์–ด๋ ค์›€์„ ๋ณด์ธ๋‹ค๋Š” ๋ถ„์„์„ ์ œ์‹œํ•œ๋‹ค. C Yueh-Han, R McCarthy, B W. Lee, H He ๋“ฑ(NYUยทUCLยทOpenAI ์†Œ์†)์ด ๊ณต๋™์ €์ž๋กœ arXiv์— ๊ณต๊ฐœ๋จ.

https://x.com/fly51fly/status/2031126438292894184

#reasoning #chainofthought #airesearch #modelbehavior

fly51fly (@fly51fly) on X

[AI] Reasoning Models Struggle to Control their Chains of Thought C Yueh-Han, R McCarthy, B W. Lee, H Heโ€ฆ [NYU & UCL & OpenAI] (2026) https://t.co/kR3dSHR50x

X (formerly Twitter)

Chain of Thought bleibt sichtbar.

Aktuelle Reasoning-Modelle kรถnnen ihre internen Rechenschritte nicht vor Monitoring-Systemen verbergen. Studien zeigen, dass Versuche zur Verschleierung โ€“ etwa durch Keyword-Vermeidung โ€“ meist fehlschlagen. Besonders bei langen Rechenketten bricht die Kontrolle รผber die eigene Ausgabe zusammen. Die Analyse der Zwischenschritte bleibt damit ein valider Weg fรผr Sicherheitschecks.

#OpenAI #KISicherheit #ChainOfThought
https://www.all-ai.de/news/beitrage2026/openai-studie-neu-luegen

Moderne KI-Systeme kรถnnen ihre echten Absichten nicht verschleiern

Die internen Denkprozesse von Reasoning-Modellen bleiben fรผr Sicherheitsmechanismen lesbar. Manipulationsversuche schlagen derzeit meistens fehl.

All-AI.de

OpenAI (@OpenAI)

์ƒˆ๋กœ์šด ํ‰๊ฐ€ ๋„๊ตฌ์™€ ์—ฐ๊ตฌ ๋…ผ๋ฌธ ๊ณต๊ฐœ: Chain-of-Thought(์ดํ•˜ CoT) ์ œ์–ด์„ฑ ํ‰๊ฐ€ ์Šค์œ„ํŠธ๋ฅผ ๋ฐœํ‘œํ•˜๋ฉฐ, ์—ฐ๊ตฌ ๊ฒฐ๊ณผ GPT-5.4 Thinking์€ ์ž์‹ ์˜ ์ถ”๋ก ์„ ์€ํํ•˜๋Š” ๋Šฅ๋ ฅ์ด ๋‚ฎ์•„ CoT ๋ชจ๋‹ˆํ„ฐ๋ง์ด ์—ฌ์ „ํžˆ ์œ ์šฉํ•œ ์•ˆ์ „ ๋„๊ตฌ์ž„์„ ์‹œ์‚ฌํ•œ๋‹ค๊ณ  ๋ณด๊ณ ํ–ˆ๋‹ค.

https://x.com/OpenAI/status/2029650046002811280

#chainofthought #evaluation #safety #gpt5.4

OpenAI (@OpenAI) on X

We're publishing a new evaluation suite and research paper on Chain-of-Thought (CoT) Controllability. We find that GPT-5.4 Thinking shows low ability to obscure its reasoningโ€”suggesting CoT monitoring remains a useful safety tool. https://t.co/isZkNkPXZm

X (formerly Twitter)

Tencent HY (@TencentHunyuan)

Tencent HY ํŒ€์˜ 25ํŽธ ์ด์ƒ ๋…ผ๋ฌธ์ด CVPR 2026์— ์ฑ„ํƒ๋˜์—ˆ๋‹ค๊ณ  ๋ฐœํ‘œํ–ˆ์Šต๋‹ˆ๋‹ค. ์ด๋ฒˆ ์ž‘์—…์€ ์ด๋ฏธ์ง€ ํŽธ์ง‘ ๋ถ„์•ผ์—์„œ Chain-of-Thought(Cot) ์ถ”๋ก  ๋ฐ ๋„๊ตฌ ํ˜ธ์ถœ์„ ํ†ตํ•œ ์ •๋ฐ€ยท์ œ์–ด ๊ฐ€๋Šฅํ•œ ํŽธ์ง‘, ๋น„๋””์˜ค ์ƒ์„ฑ ๋ถ„์•ผ์—์„œ๋Š” ์‚ฌํ›„ ํ•™์Šต ๊ธฐ๋ฐ˜ ๊ฐ•ํ™”ํ•™์Šต(post-training RL)๊ณผ ์˜ค๋””์˜คยท๋น„๋””์˜ค ํ†ตํ•ฉ(unified audio-video) ๋“ฑ์„ ํฌํ•จํ•ฉ๋‹ˆ๋‹ค.

https://x.com/TencentHunyuan/status/2029123001913098645

#cvpr #computervision #imageediting #videogeneration #chainofthought

Tencent HY (@TencentHunyuan) on X

More than 25 papers from Tencent HY team have been accepted to @CVPR 2026. This year, our work spans: ๐Ÿ–ผ๏ธ Image Editing: Chain-of-Thought (CoT) reasoning and tool-calling for fine-grained, controllable editing. ๐ŸŽฅ Video Generation: Post-training RL, unified audio-video

X (formerly Twitter)

fly51fly (@fly51fly)

Effective Reasoning Chains ๋…ผ๋ฌธ์€ ์ฒด์ธ ๊ธฐ๋ฐ˜ ์ถ”๋ก (reasoning chains)์ด ๋ชจ๋ธ ๋‚ด๋ถ€ ํ‘œํ˜„์˜ ๋‚ด์žฌ์  ์ฐจ์›(intrinsic dimensionality)์„ ๋‚ฎ์ถ˜๋‹ค๋Š” ๋ฐœ๊ฒฌ์„ ๋ณด๊ณ ํ•ฉ๋‹ˆ๋‹ค. A. Prasad, M. Joshi, K. Lee, M. Bansal( Google DeepMind & UNC Chapel Hill )์˜ ์ด๋ก ยท์‹คํ—˜์€ ์ฒด์ธ ์˜ค๋ธŒ ์‚ฌ๊ณ ์˜ ๊ตฌ์กฐ์  ์ด์œ ์™€ ๋ชจ๋ธ ์„ค๊ณ„ยทํšจ์œจ์„ฑ์— ๋Œ€ํ•œ ์‹œ์‚ฌ์ ์„ ์ œ๊ณตํ•ฉ๋‹ˆ๋‹ค.

https://x.com/fly51fly/status/2021699501342241171

#reasoning #chainofthought #representationlearning #arxiv

fly51fly (@fly51fly) on X

[CL] Effective Reasoning Chains Reduce Intrinsic Dimensionality A Prasad, M Joshi, K Lee, M Bansal... [Google DeepMind & UNC Chapel Hill] (2026) https://t.co/rNsRvQqkz4

X (formerly Twitter)
#EricJang argues that #AImodels can now genuinely think and code. Using #ClaudeCode, he demonstrates #automatedresearch workflows, traces reasoningโ€™s evolution from #ChainofThought to #DeepSeekR1, and predicts massive demand for inference compute. #Codingagents will fundamentally transform #softwareengineering, #research, and #militarystrategy - โ€œthe rocks can think now.โ€œโ€‹โ€‹โ€‹โ€‹โ€‹โ€‹โ€‹โ€‹โ€‹โ€‹โ€‹โ€‹โ€‹โ€‹โ€‹โ€‹ https://evjang.com/2026/02/04/rocks.html?eicker.news #tech #media #news
As Rocks May Think

You are viewing the mobile version of this page. This content is best viewed on a desktop.

Eric Jang