mlsanityData to 14 Oct 2026

Task Ambiguity in Humans and Language Models

arXiv 2212.10711 ↗First posted 20 Dec 2022cs.CL cs.LGcited by 28

Not on the map. Why?

Abstract (arXiv)

Language models have recently achieved strong performance across a wide range of NLP benchmarks. However, unlike benchmarks, real world tasks are often poorly specified, and agents must deduce the user's intended behavior from a combination of context, instructions, and examples. We investigate how both humans and models behave in the face of such task ambiguity by proposing AmbiBench, a new benchmark of six ambiguously-specified classification tasks. We evaluate humans and models on AmbiBench by seeing how well they identify the intended task using 1) instructions with varying degrees of ambiguity, and 2) different numbers of labeled examples. We find that the combination of model scaling (to 175B parameters) and training with human feedback data enables models to approach or exceed the accuracy of human participants across tasks, but that either one alone is not sufficient. In addition, we show how to dramatically improve the accuracy of language models trained without large-scale human feedback training by finetuning on a small number of ambiguous in-context examples, providing a promising direction for teaching models to generalize well in the face of ambiguity.

Key facts. Cited by 28 papers in the map (data through 14 Oct 2026). Builds on 35 of its 51 references that are in the map. First posted on arXiv 20 Dec 2022.

Builds on

35 of 51 references.

References · 35

  1. 2020Language Models are Few-Shot Learnerscited by 32,419
  2. 2022Training language models to follow instructions with human feedbackcited by 15,971
  3. 2022Flamingo: a Visual Language Model for Few-Shot Learningcited by 4,386
  4. 2021On the Opportunities and Risks of Foundation Modelscited by 4,033
  5. 2021Finetuned Language Models Are Zero-Shot Learnerscited by 3,780
  6. 2022Training a Helpful and Harmless Assistant with Reinforcement Learning from Human Feedbackcited by 3,720
  7. 2022Training Compute-Optimal Large Language Modelscited by 3,019
  8. 2021Pre-train, Prompt, and Predict: A Systematic Survey of Prompting Methods in Natural Language Processingcited by 2,234
  9. 2022Beyond the Imitation Game: Quantifying and extrapolating the capabilities of language modelscited by 1,998
  10. 2020Shortcut Learning in Deep Neural Networkscited by 1,969
  11. 2021Multitask Prompted Training Enables Zero-Shot Task Generalizationcited by 1,470
  12. 2022Rethinking the Role of Demonstrations: What Makes In-Context Learning Work?cited by 1,434
23 more
  1. 2021Calibrate Before Use: Improving Few-Shot Performance of Language Modelscited by 1,392
  2. 2021What Makes Good In-Context Examples for GPT-3?cited by 1,195
  3. 2021Fantastically Ordered Prompts and Where to Find Them: Overcoming Few-Shot Prompt Order Sensitivitycited by 1,135
  4. 2021An Explanation of In-context Learning as Implicit Bayesian Inferencecited by 938
  5. 2022A Generalist Agentcited by 808
  6. 2018Annotation Artifacts in Natural Language Inference Datacited by 671
  7. 2021MetaICL: Learning to Learn In Contextcited by 454
  8. 2020AmbigQA: Answering Ambiguous Open-domain Questionscited by 390
  9. 2021Do Prompt-Based Models Really Understand the Meaning of their Prompts?cited by 326
  10. 2018Probabilistic Model-Agnostic Meta-Learningcited by 256
  11. 2019Evaluating Gender Bias in Machine Translationcited by 249
  12. 2021Meta-learning via Language Model In-context Tuningcited by 164
  13. 2021Understanding the Capabilities, Limitations, and Societal Impact of Large Language Modelscited by 160
  14. 2021SituatedQA: Incorporating Extra-Linguistic Contexts into QAcited by 145
  15. 2019Asking Clarifying Questions in Open-Domain Information-Seeking Conversationscited by 139
  16. 2021Adapting Language Models for Zero-shot Learning by Meta-tuning on Dataset and Prompt Collectionscited by 128
  17. 2022Ground-Truth Labels Matter: A Deeper Look into Input-Label Demonstrationscited by 116
  18. 2021Building and Evaluating Open-Domain Dialogue Corpora with Clarifying Questionscited by 63
  19. 2020Language Through a Prism: A Spectral Approach for Multiscale Language Representationscited by 39
  20. 2022Active Learning Helps Pretrained Models Learn the Intended Taskcited by 33
  21. 2021ConditionalQA: A Complex Reading Comprehension Dataset with Conditional Answerscited by 29
  22. 2022INSCIT: Information-Seeking Conversations with Mixed-Initiative Interactionscited by 27
  23. 2022Feature Dropout: Revisiting the Role of Augmentations in Contrastive Learningcited by 9

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2023: 12; 2024: 8; 2025: 1; 2026: 7.

28 papers.

Sort:

By area:

  1. 2023Siren's Song in the AI Ocean: A Survey on Hallucination in Large Language Modelscited by 698
  2. 2023TheoremQA: A Theorem-driven Question Answering datasetcited by 192
  3. 2023Understanding Catastrophic Forgetting in Language Models via Implicit Inferencecited by 144
  4. 2023We're Afraid Language Models Aren't Modeling Ambiguitycited by 126
  5. 2024A Survey on Uncertainty Quantification of Large Language Models: Taxonomy, Open Research Challenges, and Future Directionscited by 121
  6. 2023Decomposing Uncertainty for Large Language Models through Input Clarification Ensemblingcited by 114
  7. 2024STaR-GATE: Teaching Language Models to Ask Clarifying Questionscited by 104
  8. 2023Eliciting Human Preferences with Language Modelscited by 92
  9. 2023Measuring Inductive Biases of In-Context Learning with Underspecified Demonstrationscited by 49
  10. 2023Can Large Language Models Capture Dissenting Human Voices?cited by 33
  11. 2024Introspective Planning: Aligning Robots' Uncertainty with Inherent Task Ambiguitycited by 27
  12. 2023It Takes Two to Tango: Navigating Conceptualizations of NLP Tasks and Measurements of Performancecited by 26
16 more
  1. 2024Dialogue Action Tokens: Steering Language Models in Goal-Directed Dialogue with a Multi-Turn Plannercited by 13
  2. 2024Do LLMs Understand Ambiguity in Text? A Case Study in Open-world Question Answeringcited by 13
  3. 2023Self-Consistency of Large Language Models under Ambiguitycited by 12
  4. 2023Social Contract AI: Aligning AI Assistants with Implicit Group Normscited by 9
  5. 2024A Survey on Responsible Generative AI: What to Generate and What Notcited by 8
  6. 2023Lending Interaction Wings to Recommender Systems with Conversational Agentscited by 6
  7. 2024Towards Unified Alignment Between Agents, Humans, and Environmentcited by 5
  8. 2025Towards Reliable LLM-based Robot Planning via Combined Uncertainty Estimationcited by 4
  9. 2026Ask Early, Ask Late, Ask Right: When Does Clarification Timing Matter for Long-Horizon Agents?cited by 4
  10. 2024Sensitivity of Generative VLMs to Semantically and Lexically Altered Promptscited by 2
  11. 2026ACCORD: Action-Conditioned Contextual Grounding for Language Agentscited by 0
  12. 2026Expectation Alignment of Language Models for Real-World User Expectationscited by 0
  13. 2026Where did the ambiguity go? Examining how multimodal models interpret polysemous wordscited by 0
  14. 2026Expert-Guided g-computation with Large Language Models for Estimating Causal Effects on Timings: Applications to Hospital Quality Improvementcited by 0
  15. 2026A knowledge-guided agentic framework for mitigating patient-context ambiguity in health queriescited by 0
  16. 2026XYEval: Agents say yes to bad advicecited by 0