Why Language Models Hallucinate
Adam Tauman Kalai et al.
arXiv:2509.04664
In short
The authors argue hallucinations are ordinary statistical errors: pre-training cannot always tell true from plausible-but-false statements. They persist because most benchmarks score an “I don’t know” as zero, so models are rewarded for confident guessing, and the fix is to change how evaluations are graded.
Why it matters
It reframes hallucination as an incentive problem baked into how we measure models, not a mystery.
Read first
The 4 Field Guide ideas this paper leans on.
Starting from scratch? The full route 22 ideas · basics first
- Natural Language Processing ✓ understood
The field of AI that lets computers read, interpret, translate and generate human language, from spam filters and search to chatbots.
- Token ✓ understood
The unit of text a language model reads and writes: a word, part of a word, or a character, drawn from the model's fixed vocabulary.
- Tokenization ✓ understood
Splitting text into tokens, usually subword pieces, and mapping each to an integer ID so a language model can process it.
- Language Modeling · read first ✓ understood
Assigning probabilities to sequences of text, in practice by predicting each next token from the ones before it; the objective behind large language models.
- Machine Learning ✓ understood
Building systems that learn patterns from data instead of following hand-written rules, getting better at a task as they see more examples.
- Supervised Learning ✓ understood
Learning from examples paired with the correct answer, so a model can predict answers for new inputs it hasn't seen.
- Classification · read first ✓ understood
A supervised learning task where the model assigns each input to one of a fixed set of categories, such as spam or not spam.
- Dataset ✓ understood
A collection of data examples used for training, validating, or testing machine learning models.
- Benchmark · read first ✓ understood
A standardized dataset and task used to compare model performance across different approaches (ImageNet, GLUE, SuperGLUE).
- Training ✓ understood
The process of fitting a model to data by repeatedly measuring how wrong its outputs are and adjusting its parameters to reduce that error.
- Unsupervised Learning ✓ understood
Learning from unlabeled data to discover hidden patterns, structures, or relationships without explicit target outputs.
- Self-Supervised Learning ✓ understood
Training on unlabeled data by making the data supply its own answers: hide or alter part of an input and have the model predict or match it.
- Pre-training ✓ understood
Training a model on a large dataset (often self-supervised) before fine-tuning on specific tasks, enabling transfer learning.
- Feature ✓ understood
A single measurable property of an example, such as a house's floor area or how many links an email contains, used as an input to a model.
- Neural Network ✓ understood
A model built from layers of simple units, each taking a weighted sum of its inputs and applying a nonlinearity; training adjusts the weights so the whole stack maps inputs to useful outputs.
- Deep Learning ✓ understood
A subset of machine learning that uses neural networks with multiple layers (deep neural networks) to learn hierarchical representations of data.
- Representation Learning ✓ understood
Letting a model learn its own internal description of the data, such as embeddings or layered visual features, instead of relying on features designed by hand.
- Embedding ✓ understood
A list of numbers (a vector) that represents a word, sentence, image or other item, learned so that similar items end up close together.
- Attention Mechanism ✓ understood
A technique that lets a neural network weigh every part of its input when producing each output, focusing on the parts most relevant at that step.
- Transformer ✓ understood
A neural network architecture, introduced in 2017, built from stacked self-attention and feed-forward layers; the basis of nearly every modern large language model.
- Large Language Model ✓ understood
A neural network, almost always a transformer, trained on vast amounts of text to predict the next token, which lets it write, answer, summarize and follow instructions.
- Hallucination · read first ✓ understood
When a language model produces fluent, confident output that is false or unsupported by its sources, such as invented facts, quotes or citations.
In the frontier
- Rank
- #12 of 100
- Citations
- 282
- as of Aug 9, 2026
- Published
- Sep 2025
Topics: Interpretability and analysis , Safety and alignment
Selection: 1kpapers.com by Together AI, most-cited as of Aug 9, 2026