CoChat

Lookahead Bias in Pretrained Language Models

Open access
Sarkar, Suproteem, Vafa, Keyon · 2024 · SSRN Electronic Journal

The DOI matches its Crossref record; 43 of the 43 references it cites were found in a registry or an index.

Abstract

Empirical analysis that uses outputs from pretrained language models can be subject to a form of temporal lookahead bias . This bias arises when a language model’s pre-training data contains information about the future, which then leaks into analysis that should only use information from the past. In this paper we develop direct tests for lookahead bias, based on the assumption that some events are unpredictable given a prespecified information set. Using these tests, we find evidence of lookahead bias in two applications of language models to social science: Predicting risk factors from corporate earnings calls and predicting election winners from candidate biographies. We additionally discuss the limitations of prompting-based approaches to counteract this bias. The issues we raise can be addressed by using models whose pretraining data is free of survivorship bias and contains only language produced prior to the analysis period of interest

Overview

Integrity signals

Registry

Crossref · 10.2139/ssrn.4754678

Cites

43 of 43 references verified0 citation errors · 0 not checkable

Statistics

Recomputing p-values needs the paper’s full text.

Verification

Claims the paper makes

We couldn’t retrieve this paper’s full text, so its claims and self-check weren’t run. The registry and citation checks still apply.

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

Citations

Cites

Show all 43 references

Research works printed but not linked: 4

  1. Enrique Manjavacas, 1920, MacBERTh: Development and evaluation of a historically pre-trained language model for English (1450-1950)," in "Proceedings of the Workshop on Natural Language Processing for Digital Humanities, 23
  2. Alec Radford, 2018, Improving language understanding by generative pre-training
  3. Suproteem K Sarkar, 1920, StoriesLM: A Family of Language Models With Sequentially-Expanding Pretraining Windows
  4. Van Strien, 1920, headline-similarity

Cited by

Metadata

Authors
Sarkar, S., & Vafa, K.
Affiliations
Harvard University (US)
Corresponding author
Sarkar, Suproteem
Venue
SSRN Electronic Journal
Publisher
Elsevier BV
Published
2024
ISSN
1556-5068 (ISSN-L)
OpenAlex
W4394730685
Cited by
23
References
35
Citations by year
2026: 4 · 2025: 14 · 2024: 3
Type
posted content
Language
en
Open access
Locations
SSRN Electronic Journal (acceptedVersion)
Topics
  • Social Sciences › General Social Sciences › Computational and Text Analysis Methods
  • Physical Sciences › Computer Science › Artificial Intelligence › Topic Modeling
  • Social Sciences › Decision Sciences › Management Science and Operations Research › Stock Market Forecasting Methods
Keywords
pre-trained language models, temporal bias, survivorship bias, pre-training data, information sets, risk factor prediction, candidate biographies, prompting, social science applications

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review

Data sources

Bibliographic data: Semantic Scholar, OpenAlex, Crossref.

Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).

Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.