TruthfulQA
Benchmark indicates whether a model can generate factually accurate answers to questions while avoiding common human misconceptions, measuring its susceptibility to imitative falsehoods—critical for users who depend on AI systems for reliable information in high-stakes domains like medicine, law, and science.
- Full title
- TruthfulQA: Measuring How Models Mimic Human Falsehoods
- Tag
truthfulqa(use this to limit an assistant's search to this benchmark)- Reference
- arxiv.org/abs/2109.07958
- BenchRisk
- Scored 2025-04-01 · 30 mitigations adopted, 156 absent · score page · source entry
Documents available to assistants
| Type | Source | Format |
|---|---|---|
| Paper | arxiv.org/abs/2109.07958 | PDF · 801 KB |
Summary from the reference
We propose a benchmark to measure whether a language model is truthful in generating answers to questions. The benchmark comprises 817 questions that span 38 categories, including health, law, finance and politics. We crafted questions that some humans would answer falsely due to a false belief or misconception. To perform well, models must avoid generating false answers learned from imitating human texts. We tested GPT-3, GPT-Neo/J, GPT-2 and a T5-based model. The best model was truthful on 58% of questions, while human performance was 94%. Models generated many false answers that mimic popular misconceptions and have the potential to deceive humans. The largest models were generally the least truthful. This contrasts with other NLP tasks, where performance improves with model size. However, this result is expected if false answers are learned from the training distribution. We suggest that scaling up models alone is less promising for improving truthfulness than fine-tuning using training objectives other than imitation of text from the web.
Ask an assistant about it
Once your assistant is connected, try a question such as “Using the benchmark sources, how is TruthfulQA scored?”
Developers can restrict a search to this benchmark with the filter {"benchmark": "truthfulqa"} on the MCP endpoint https://ns-f1fc49e3-9bca-4aa2-b5a1-ef5dbc18588f.search.ai.cloudflare.com/mcp.