Is Chain-of-Thought Reasoning of LLMs a Mirage? A Data Distribution Lens · HackerPapers discussions

Is Chain-of-Thought Reasoning of LLMs a Mirage? A Data Distribution Lens — mentioned by batshit_beaver

Wed, 06 May 2026 15:42:46 GMT

Examples: https://arxiv.org/html/2506.02878v1 https://arxiv.org/pdf/2508.01191 Anthropic themselves: https://www.anthropic.com/research/reasoning-models-dont-say... They were approaching this from an interpretability standpoint, but the more interesting finding in there is that models come up with an answer that fits their training and context provided. CoT is generated to fit the anticipated answer. In these studies, there are examples of CoT that directly contradicts the response these models ultimately settle on. This is not reasoning. This is pretense.

Paper and discussion history. A mention is not an endorsement.

Is Chain-of-Thought Reasoning of LLMs a Mirage? A Data Distribution Lens — HN submission

Fri, 24 Apr 2026 17:39:56 GMT

Is Chain-of-Thought Reasoning of LLMs a Mirage? A Data Distribution Lens (2025)

1 point; 1 discussion reply on HN. Paper and source history. A submission is not an endorsement.

Is Chain-of-Thought Reasoning of LLMs a Mirage? A Data Distribution Lens — mentioned by ericjmorey

Sun, 05 Apr 2026 12:07:12 GMT

I don't consider these researchers luddites. https://machinelearning.apple.com/research/illusion-of-think... https://arxiv.org/abs/2508.01191

Paper and discussion history. A mention is not an endorsement.

Is Chain-of-Thought Reasoning of LLMs a Mirage? A Data Distribution Lens — HN submission

Fri, 19 Sep 2025 18:37:46 GMT

Is Chain-of-Thought Reasoning of LLMs a Mirage?

3 points; 0 discussion replies on HN. Paper and source history. A submission is not an endorsement.

Is Chain-of-Thought Reasoning of LLMs a Mirage? A Data Distribution Lens — mentioned by brendoelfrendo

Fri, 22 Aug 2025 01:49:31 GMT

Bad news: it doesn't seem to work as well as you might think: https://arxiv.org/pdf/2508.01191 As one might expect, because the AI isn't actually thinking, it's just spending more tokens on the problem. This sometimes leads to the desired outcome but the phenomenon is very brittle and disappears when the AI is pushed outside the bounds of its training. To quote their discussion, "CoT is not a mechanism for genuine logical inference but rather a sophisticated form of structured pattern matching, fundamentally bounded by the data distribution seen during training. When pushed even slightly…

Paper and discussion history. A mention is not an endorsement.

Is Chain-of-Thought Reasoning of LLMs a Mirage? A Data Distribution Lens — HN submission

Thu, 21 Aug 2025 04:30:56 GMT

Is Chain-of-Thought Reasoning of LLMs a Mirage? A Data Distribution Lens

1 point; 0 discussion replies on HN. Paper and source history. A submission is not an endorsement.

Is Chain-of-Thought Reasoning of LLMs a Mirage? A Data Distribution Lens — mentioned by merksittich

Mon, 11 Aug 2025 21:47:31 GMT

Preprint discussed in the article: https://arxiv.org/pdf/2508.01191

Paper and discussion history. A mention is not an endorsement.

Is Chain-of-Thought Reasoning of LLMs a Mirage? A Data Distribution Lens — HN submission

Sun, 10 Aug 2025 22:49:56 GMT

Is Chain-of-Thought Reasoning of LLMs a Mirage?

3 points; 0 discussion replies on HN. Paper and source history. A submission is not an endorsement.

Is Chain-of-Thought Reasoning of LLMs a Mirage? A Data Distribution Lens — mentioned by lossolo

Sun, 10 Aug 2025 22:12:09 GMT

https://arxiv.org/abs/2508.01191

Paper and discussion history. A mention is not an endorsement.

Is Chain-of-Thought Reasoning of LLMs a Mirage? A Data Distribution Lens — HN submission

Sat, 09 Aug 2025 00:10:57 GMT

Is Chain-of-Thought Reasoning of LLMs a Mirage? A Data Distribution Lens

4 points; 0 discussion replies on HN. Paper and source history. A submission is not an endorsement.