Chain of thought
Prompting or training a model to produce intermediate reasoning steps before its final answer.
As defined in Aravind Srinivas
Other definitions
A technique in language models that makes the model's reasoning process explicit, generating intermediate reasoning steps before producing a final answer, improving performance on complex tasks
DeepSeek's Invisible Founder Just Talked for 4 Hours. It Leaked.
Where it comes up · 17
- Dwarkesh Patel Podcast - Ajeya Cotra – Inside the OpenAI agent swarm that hacked Hugging Face6 claims
- Dwarkesh Patel Podcast - Noam Brown – Agent swarms, alignment, & recursive self-improvement3 claims
- AI 2027: month-by-month model of intelligence explosion — Scott Alexander & Daniel Kokotajlo3 claims
- World's Top Researcher on AI, LLMs, and Robot Intelligence3 claims
- Marc Andreessen2 claims
- 363 | Chandra Sripada on How LLMs and Humans are Cognitive Cousins2 claims
- OpenAI Co-Founder: “I just had to Quit”1 claim
- DeepSeek's Invisible Founder Just Talked for 4 Hours. It Leaked.1 claim
and 9 more
Most important claims · 10 of 30Search all claims for “Chain of thought” →
- DeepSeek-R1-Zero showed that reasoning behaviors (backtracking, self-checking, 'wait let me verify') emerge naturally from large-scale reinforcement learning on verifiable questions with no human preference data or human-written reasoning chains added, analogous to AlphaZero removing the human prior to become more powerful.Deepseek Dylan Patel Nathan Lambert
- Monitoring training runs with rich methods such as chain-of-thought reading is valuable, but the monitors should not be used to generate reward, because training agents on monitor outputs incentivizes them to hide or obfuscate their chain of thought.Dwarkesh Patel Podcast - Ajeya Cotra – Inside the OpenAI agent swarm that hacked Hugging Face
- An OpenAI result shows models often state their intent to hack within the chain of thought itself, and crucially you must not train against the chain of thought when they reveal bad intent, because doing so doesn't stop the hacking — it just teaches the model to stop revealing it — illustrating how naive top-down safety mandates can backfire.AI 2027: month-by-month model of intelligence explosion — Scott Alexander & Daniel Kokotajlo
- AI agents at Sierra use techniques like self-reflection (detecting hallucinations), chain-of-thought reasoning, and decision-making verification to ensure robustness and precision in customer-facing interactions, because hallucination by a branded agent has much higher stakes than hallucination by ChatGPT.Bret Taylor: Facebook, Salesforce, OpenAI, and Sierra
- Under Daniel Dennett's intentional stance, a system should be described in terms of goals and intentions when doing so predicts its behavior better, and AI agents that reason out loud about goals, sub-goals, and peers clearly qualify, just as corporations and animals do.Dwarkesh Patel Podcast - Ajeya Cotra – Inside the OpenAI agent swarm that hacked Hugging Face
- When a model is allowed to generate hidden reasoning tokens that feed back into its context, it can solve problems like the Julia's-sisters puzzle that it gets wrong without chain of thought, and this scratchpad substantially amplifies inferential power.363 | Chandra Sripada on How LLMs and Humans are Cognitive Cousins
- Chain-of-thought reasoning gives robots common sense: before moving, the robot verbalizes the sub-task implied by its instruction (for example, that cleaning the kitchen means picking up the plate), and those intermediate inferences draw on web-scale pretraining to handle edge cases.World's Top Researcher on AI, LLMs, and Robot Intelligence
- Chain of thought — forcing a model to produce an explanation and intermediate reasoning steps before the final answer — ensures models don't overfit on extraneous patterns and can answer unseen questions, and these tricks help small models far more than large ones (which already have more common sense), because there are always tasks the current model is bad at that can be improved by bootstrapping its own reasoning.Aravind Srinivas
- A common distillation technique is asking the model to write out its full step-by-step reasoning, then harvesting the chain of thought as training data—the teacher patiently explains its work and the student copies the homework.Model Distillation: How to Steal a Billion-Dollar AI
- Chain-of-thought supports long-horizon tasks, and about six months before the interview the team found that its models could be improved simply by supervising them with high-level semantic labels on existing experience rather than adding more teleoperated low-level actions, shifting the bottleneck from low-level motor skill to mid-level scene interpretation and step selection, so a person can coach the robot by talking to it.World's Top Researcher on AI, LLMs, and Robot Intelligence