Làkk AI GUIDE

When Chain-of-Thought Hurts

Chain-of-thought prompting asks a model to produce intermediate reasoning before its answer.

Déglul tegtal biiIA-nettali · ~3 min

Ci xët wii3 simili jàng
  1. Résumé
  2. Plongeur bu xóot
  3. Doxal ci àdduna dëgg
  4. The Future of When Chain-of-Thought Hurts
  5. Teg ko ci jëf
  6. Weyal di banneexu
  7. Laaj yi ñuy faral di laaj

Résumé

It can help on some multi-step tasks, but research shows effects vary by task and model; for some tasks, extra deliberation can reduce performance, add latency, or produce no benefit.

Plongeur bu xóot

Chain-of-thought (CoT) prompting asks a model to write intermediate steps before its final answer. It became a prominent technique after research reported benefits on selected multi-step reasoning benchmarks. But “ask for reasoning” is not a universal improvement. A 2025 ICML paper evaluated six tasks drawn from psychological studies where deliberation can hurt human performance. The researchers found significant CoT-related drops for state-of-the-art models on three tasks, while results on the other tasks were mixed. That study gives evidence that performance can fall in particular settings; it does not show that CoT generally harms models or identify one rule that predicts every task. Extra written steps also have a practical cost: they use output space and may increase latency. More text can introduce an unsupported assumption that the model then carries into its answer. An explanation should not be mistaken for a faithful record of the internal process or proof that a conclusion is correct. For reasoning models, the appropriate prompting advice can differ. OpenAI’s current API guide, for example, recommends avoiding “think step by step” instructions for its reasoning models, while its model-specific guidance for some non-reasoning models may discuss other prompting approaches. Follow the documentation for the model being tested. Choose based on evidence from the task. Compare direct and CoT variants on the same representative examples, use a predefined scoring rule, and include latency or token limits if they matter to the application. Keep an approach only when it improves the required outcomes without unacceptable costs. For high-stakes decisions, independent checks and expert review matter more than whether the model displays an explanation. Avoid assuming that a longer rationale is inherently more transparent, reliable, or safe.

Doxal ci àdduna dëgg

A team compares direct answers with step-by-step prompting on a set of its own short classification tasks before adopting a default.

A low-latency service tests whether extra explanation changes accuracy enough to justify the additional response time.

A researcher uses a published evaluation to identify task types where a specific model’s performance drops under chain-of-thought prompting.

A prompt author testing an OpenAI reasoning model follows the provider’s recommendation not to request a chain of thought, then evaluates the response against task criteria.

The Future of When Chain-of-Thought Hurts

Research will continue to identify which task and model combinations benefit from explicit intermediate text and which do not. Reasoning-capable products may also expose model-specific controls that make older prompt recipes less relevant. Teams should keep evaluation results tied to versions and data, and re-run comparisons when either changes. The stable principle is to test the prompt technique against the task rather than treating it as a universal default. That keeps findings tied to actual use rather than broad speculation.

Teg ko ci jëf

Lu tax mu am solo

  • Gaawaay ak yaatuwaay. Liggéeyukaay yi ci làkk yi mën nañu gëna gaaw te duñu yàq deggoo gi.
  • Dugg ak yegg. Dafay yaatal jëfandikoo gi ci làkk yi ak ci anam yi ñuy jokkoo.
  • dogal yu gëna leer. Ekip yi mën nañu gëna yàgg ci àtte ci jamono ji otomatisation di liggéey ci baamtu.

Moytul

  • Lépp lu jaarul yoon mën na dugg ci rapoor yi, jàppale ci liggéey bi, wala ci njariñu gëstu bi.
  • Sensibilite bu gaaw mën na jur njariñ yu wuute ci laajte yu noonu mel.
  • Done yu am solo mën nañu feeñ sudee seytu jëfandikoo gi néew doole.

Jéego yi ci topp

  1. Mandargal formaa génne gi, melokaan bi, ak standard kalite yi laata ngay dugal ko.
  2. Tontu yu am solo ak balluwaay yu wóor saa yu dëggu bi di am solo.
  3. Fexeel am barabu xool nit ñi ngir am njariñ yu am solo.
  4. Toppal anami gacce yi ak di faral di tàggataat ay laaj wala def-liggéey.

Weyal di banneexu

Bulletin bu amul fayda

Jàngal xibaar AI bu bes

Ñatti jaar-jaari IA yuñ firndeel bis bu nekk ci ayu-bis bi, ñu bind leen ci Àngle bu leer. Gratis ba fàww, amul yëgle.

Benn email bis bu nekk ci ayu-bis bi. Benn klik nga dindi abonemaa bi. Dunu musa jaay wala séddoo sa adres.

Nattal sa bopp

Take the When Chain-of-Thought Hurts quiz

Feedback ci saasi ci bépp tontu, ak sertifikaa buñ mëna séddoo ak ID buñ mëna saytu soo jàllee ci njàng mi.

Tambalil quiz

Jàppale njàngum IA te doo fay. AI Understanding mbootaay la buy def te Yàlla tax — amul yëgle, amul paywall, musul am. Defal ab sarax

Laaj yi ñuy faral di laaj

What is When Chain-of-Thought Hurts?

Chain-of-thought prompting asks a model to produce intermediate reasoning before its answer. It can help on some multi-step tasks, but research shows effects vary by task and model; for some tasks, extra deliberation can reduce performance, add latency, or produce no benefit.

What does chain-of-thought prompting ask a model to produce?

The guide defines CoT as asking for intermediate reasoning before the final answer.

What did the 2025 ICML study report across its six selected tasks?

The paper reports significant drops for models on three of six tasks and mixed results on the rest.

What does that study establish about CoT across all AI tasks?

The guide stresses that the paper’s six-task finding is bounded and does not prove general harm.

Why might an explicit rationale add operational cost?

The guide notes that written steps consume output space and may increase response time.

How should a model-generated explanation be treated as evidence?

The guide warns against treating an explanation as proof of correctness or faithful internal reasoning.