Language models are better than humans at next-token prediction

TL;DR

This paper demonstrates that language models outperform humans in next-token prediction tasks, challenging the perception of their sub-human capabilities in natural language processing.

Contribution

The study provides the first direct comparison showing that language models surpass humans in next-token prediction accuracy and perplexity.

Findings

Language models outperform humans in next-token prediction.

Humans are consistently worse than small language models like GPT-3-Ada.

Results challenge assumptions about language models' capabilities.

Abstract

Current language models are considered to have sub-human capabilities at natural language tasks like question-answering or writing code. However, language models are not trained to perform well at these tasks, they are trained to accurately predict the next token given previous tokes in tokenized text. It is not clear whether language models are better or worse than humans at next token prediction. To try to answer this question, we performed two distinct experiments to directly compare humans and language models on this front: one measuring top-1 accuracy and the other measuring perplexity. In both experiments, we find humans to be consistently \emph{worse} than even relatively small language models like GPT3-Ada at next-token prediction.