Co-Writing with Opinionated Language Models Affects Users’ Views
Informations
Type:
inproceedings
Auteurs:
Jakesch, Maurice and Bhat, Advait and Buschek, Daniel and Zalmanson, Lior and Naaman, Mor
Pertinence:
Haute
Référence:
Doi:
10.1145/3544548.3581196
Mots-clés:
GPT-3, risks of large language models, Co-writing, opinion change
Url:
https://doi.org/10.1145/3544548.3581196
Date de publication:
04/2023
Résumé:
Ajout d'un copilot pour écrire dans un éditeur de texte.
Abstract:
If large language models like GPT-3 preferably produce a particular point of view, they may influence people’s opinions on an unknown scale. This study investigates whether a language-model-powered writing assistant that generates some opinions more often than others impacts what users write – and what they think. In an online experiment, we asked participants (N=1,506) to write a post discussing whether social media is good for society. Treatment group participants used a language-model-powered writing assistant configured to argue that social media is good or bad for society. Participants then completed a social media attitude survey, and independent judges (N=500) evaluated the opinions expressed in their writing. Using the opinionated language model affected the opinions expressed in participants’ writing and shifted their opinions in the subsequent attitude survey. We discuss the wider implications of our results and argue that the opinions built into AI language technologies need to be monitored and engineered more carefully.
Pdf:
Lien pdf
Références
4 articles
Titre Type Pertinence Auteurs Date Publication Références Citations Actions
Wordcraft: Story Writing With Large Language Models inproceedings Haute Yuan, Ann and Coenen, Andy and Reif, Emily and Ippolito, Daphne 03/2022 3 6
Where to Hide a Stolen Elephant: Leaps in Creative Writing with Multimodal Machine Intelligence article Haute Singh, Nikhil and Bernal, Guillermo and Savchenko, Daria and Glassman, Elena L. 02/2022 0 4
AI as an Active Writer: Interaction strategies with generated text in human-AI collaborative fiction... inproceedings Haute Yang, Daijin and Zhou, Yanpeng and Zhang, Zhiyuan and Li, Toby Jia-Jun and LC, Ray 12/2021 0 1
Attention is all you need article Haute Vaswani, Ashish and Shazeer, Noam and Parmar, Niki and Uszkoreit, Jakob and Jones, Llion and Gomez,... 12/2016 0 2
Citations
0 articles
Titre Type Pertinence Auteurs Date Publication Références Citations Actions
Pas encore d'article
Mots-clés
4 mots-clés
Nom Nombre d'articles Actions
risks of large language models 1
GPT-3 1
opinion change 1
Co-writing 1
Auteurs
6 auteurs
Nom Nombre d'articles Actions
Maurice and Bhat 1
Lior and Naaman 1
Jakesch 1
Daniel and Zalmanson 1
Advait and Buschek 1
Mor 1