Correlated Errors in Large Language Models
Fuente:
arXiv
Salvato in:
| Autori principali: | , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2025
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866910996581318656 |
|---|---|
| author | Kim, Elliot Garg, Avi Peng, Kenny Garg, Nikhil |
| author_facet | Kim, Elliot Garg, Avi Peng, Kenny Garg, Nikhil |
| contents | Diversity in training data, architecture, and providers is assumed to mitigate homogeneity in LLMs. However, we lack empirical evidence on whether different LLMs differ meaningfully. We conduct a large-scale empirical evaluation on over 350 LLMs overall, using two popular leaderboards and a resume-screening task. We find substantial correlation in model errors -- on one leaderboard dataset, models agree 60% of the time when both models err. We identify factors driving model correlation, including shared architectures and providers. Crucially, however, larger and more accurate models have highly correlated errors, even with distinct architectures and providers. Finally, we show the effects of correlation in two downstream tasks: LLM-as-judge evaluation and hiring -- the latter reflecting theoretical predictions regarding algorithmic monoculture. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2506_07962 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Correlated Errors in Large Language Models Kim, Elliot Garg, Avi Peng, Kenny Garg, Nikhil Computation and Language Artificial Intelligence Computers and Society Machine Learning Diversity in training data, architecture, and providers is assumed to mitigate homogeneity in LLMs. However, we lack empirical evidence on whether different LLMs differ meaningfully. We conduct a large-scale empirical evaluation on over 350 LLMs overall, using two popular leaderboards and a resume-screening task. We find substantial correlation in model errors -- on one leaderboard dataset, models agree 60% of the time when both models err. We identify factors driving model correlation, including shared architectures and providers. Crucially, however, larger and more accurate models have highly correlated errors, even with distinct architectures and providers. Finally, we show the effects of correlation in two downstream tasks: LLM-as-judge evaluation and hiring -- the latter reflecting theoretical predictions regarding algorithmic monoculture. |
| title | Correlated Errors in Large Language Models |
| topic | Computation and Language Artificial Intelligence Computers and Society Machine Learning |
| url | https://arxiv.org/abs/2506.07962 |