Pith. sign in

REVIEW 6 cited by

Recipes for Safety in Open-domain Chatbots

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 2010.07079 v3 pith:XTUU24TS submitted 2020-10-14 cs.CL cs.AI

classification cs.CLcs.AI
keywords modelsgenerativehumanmethodsopen-domainsafersafetyanalyzing
verification ladder T0 review T1 audit T2 compute T3 formal
0 comments
read the original abstract

Models trained on large unlabeled corpora of human interactions will learn patterns and mimic behaviors therein, which include offensive or otherwise toxic behavior and unwanted biases. We investigate a variety of methods to mitigate these issues in the context of open-domain generative dialogue models. We introduce a new human-and-model-in-the-loop framework for both training safer models and for evaluating them, as well as a novel method to distill safety considerations inside generative models without the use of an external classifier at deployment time. We conduct experiments comparing these methods and find our new techniques are (i) safer than existing models as measured by automatic and human evaluations while (ii) maintaining usability metrics such as engagingness relative to the state of the art. We then discuss the limitations of this work by analyzing failure cases of our models.

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 6 Pith papers

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

  1. Test-Time Detoxification without Training or Learning Anything

    cs.CL 2026-01 conditional novelty 6.0 of 10

    A few zeroth-order gradient steps on prompt embeddings, using only forward evaluations and a toxicity scorer, reduce toxic LLM outputs while preserving fluency.

  2. Reinforcement Learning from Human Feedback with High-Confidence Safety Constraints

    cs.LG 2025-06 conditional novelty 6.0 of 10

    HC-RLHF returns an aligned language model only after a held-out safety test certifies, with probability at least 1-delta, that expected harm (as judged by a learned cost model) is below a chosen threshold.

  3. Safe Inference-Time Alignment via Lagrangian Reward Augmentation

    cs.LG 2026-07 conditional novelty 5.5 of 10

    Dualizing Safe RLHF yields a one-dimensional convex calibration of λ that defines a drop-in safety-aware reward for Best-of-N and token-level inference-time decoders.

  4. A Survey on Responsible LLMs: Inherent Risk, Malicious Use, and Mitigation Strategy

    cs.AI 2025-01 conditional novelty 4.0 of 10

    A survey that organizes responsible-LLM research into five risk dimensions and four intervention phases, reviewing privacy, hallucination, value, toxicity, and jailbreak mitigation.

  5. Agentic Web: Weaving the Next Web with AI Agents

    cs.AI 2025-07 conditional novelty 3.0 of 10

    A position paper defines the Agentic Web as the next web era and proposes a three-dimensional conceptual framework for understanding and building it.

  6. From Hallucinations to Facts: Enhancing Language Models with Curated Knowledge Graphs

    cs.CL 2024-12 reject novelty 3.0 of 10

    Grounding chatbot replies with curated knowledge-graph triples raises human-rated factual consistency by roughly two points on a five-point scale, but the metric mostly rewards including the provided triple.

Pith tools