You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Artificial Artificial Artificial Intelligence: Crowd Workers Widely Use
Large Language Models for Text Production Tasks, Veniamin Veselovsky+, N/A, arXiv'23
#780
Large language models (LLMs) are remarkable data annotators. They can be usedto generate high-fidelity supervised training data, as well as survey andexperimental data. With the widespread adoption of LLMs, human gold--standardannotations are key to understanding the capabilities of LLMs and the validityof their results. However, crowdsourcing, an important, inexpensive way toobtain human annotations, may itself be impacted by LLMs, as crowd workers havefinancial incentives to use LLMs to increase their productivity and income. Toinvestigate this concern, we conducted a case study on the prevalence of LLMusage by crowd workers. We reran an abstract summarization task from theliterature on Amazon Mechanical Turk and, through a combination of keystrokedetection and synthetic text classification, estimate that 33-46% of crowdworkers used LLMs when completing the task. Although generalization to other,less LLM-friendly tasks is unclear, our results call for platforms,researchers, and crowd workers to find new ways to ensure that human dataremain human, perhaps using the methodology proposed here as a stepping stone.Code/data: https://github.com/epfl-dlab/GPTurk
AkihikoWatanabe
changed the title
あ
Artificial Artificial Artificial Intelligence: Crowd Workers Widely Use
Large Language Models for Text Production Tasks, Veniamin Veselovsky+, N/A, arXiv'23
Jul 3, 2023
URL
Affiliations
Abstract
Translation (by gpt-3.5-turbo)
コード/データ:https://github.com/epfl-dlab/GPTurk
Summary (by gpt-3.5-turbo)
The text was updated successfully, but these errors were encountered: