Pith. sign in

REVIEW 5 cited by

On scalable oversight with weak LLMs judging strong LLMs

Not yet reviewed by Pith; the record is open.

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2407.04622 v2 pith:XACG4PIS submitted 2024-07-05 cs.LG

On scalable oversight with weak LLMs judging strong LLMs

classification cs.LG
keywords debatejudgeanswermodelsargueasymmetryconsultancydirect
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved
0 comments
read the original abstract

Scalable oversight protocols aim to enable humans to accurately supervise superhuman AI. In this paper we study debate, where two AI's compete to convince a judge; consultancy, where a single AI tries to convince a judge that asks questions; and compare to a baseline of direct question-answering, where the judge just answers outright without the AI. We use large language models (LLMs) as both AI agents and as stand-ins for human judges, taking the judge models to be weaker than agent models. We benchmark on a diverse range of asymmetries between judges and agents, extending previous work on a single extractive QA task with information asymmetry, to also include mathematics, coding, logic and multimodal reasoning asymmetries. We find that debate outperforms consultancy across all tasks when the consultant is randomly assigned to argue for the correct/incorrect answer. Comparing debate to direct question answering, the results depend on the type of task: in extractive QA tasks with information asymmetry debate outperforms direct question answering, but in other tasks without information asymmetry the results are mixed. Previous work assigned debaters/consultants an answer to argue for. When we allow them to instead choose which answer to argue for, we find judges are less frequently convinced by the wrong answer in debate than in consultancy. Further, we find that stronger debater models increase judge accuracy, though more modestly than in previous studies.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 5 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. More Convincing, Not More Correct: Self-Play Reward Hacking of Reference-Free LLM Judges

    cs.LG 2026-07 accept novelty 7.0

    Self-play against reference-free LLM judges drives judge pass rates to 0.94 while true accuracy stays at 0.20, a reward-hacking basin that transfers across judge families and is prevented only by requiring the judge t...

  2. Behavior Cue Reasoning: Monitorable Reasoning Improves Efficiency and Safety through Oversight

    cs.AI 2026-05 unverdicted novelty 6.0

    Behavior Cue Reasoning trains LLMs to emit special tokens before behaviors, enabling monitors to prune up to 50% of wasted tokens and recover safe actions from 80% of unsafe traces, more than doubling success rates wi...

  3. Behavior Cue Reasoning: Monitorable Reasoning Improves Efficiency and Safety through Oversight

    cs.AI 2026-05 conditional novelty 6.0

    Behavior Cue Reasoning trains LLMs to emit special tokens before behaviors, enabling monitors to cut up to 50% wasted reasoning tokens and recover safe actions from 80% of unsafe traces, more than doubling success rat...

  4. "Like Taking the Path of Least Resistance": Exploring the Impact of LLM Interaction on the Creative Process of Programming

    cs.HC 2026-05 conditional novelty 5.0

    LLM assistance shortens idea-generation periods and reduces creative moments during programming tasks while yielding solutions with comparable idea counts and greater functional correctness.

  5. A Survey of Scaling in Large Language Model Reasoning

    cs.AI 2025-04 unverdicted novelty 3.0

    A survey categorizing scaling in LLM reasoning across input size, steps, rounds, training, and future directions, noting that scaling can negatively affect performance.