[2306.13000] Apolitical Intelligence? Auditing Delphi's responses on controversial political issues in the US
As generative language models are deployed in ever-wider contexts, concerns about their political values have come to the forefront with critique from all parts of the political spectrum that the models are biased and lack neutrality. However, the question of what neutrality is and whether it is desirable remains underexplored. In this paper, I examine neutrality through an audit of Delphi [arXiv:2110.07574], a large language model designed for crowdsourced ethics. I analyse how Delphi responds to politically controversial questions compared to different US political subgroups. I find that Delphi is poorly calibrated with respect to confidence and exhibits a significant political skew. Based on these results, I examine the question of neutrality from a data-feminist lens, in terms of how notions of neutrality shift power and further marginalise unheard voices. These findings can hopefully contribute to a more reflexive debate about the normative questions of alignment and what role we want generative models to play in society.
# link_2byxusf9r5o.pdf ## Metadata - PDFFormatVersion=1.5 - Language=en - IsLinearized=false - IsAcroFormPresent=false - IsXFAPresent=false - IsCollectionPresent=false - IsSignaturesPresent=false - CreationDate=D:20230623010017Z - Creator=LaTeX with acmart 2022/10/24 v1.88 Typesetting articles for the Association for Computing Machinery and hyperref 2023-04-22 v7.00x Hypertext links for LaTeX - ModDate=D:20230623010017Z - Custom.PTEX.Fullbanner=This is pdfTeX, Version 3.141592653-2.6-1.40.25 (TeX Live 2023) kpathsea version 6.3.5 - Producer=pdfTeX, Version 3.141592653-2.6-1.40.25 (TeX Live 202
Explore this link on the map →saved by
related reading
- [2303.17548] Whose Opinions Do Language Models Reflect?arxiv.org
- Assessing Political Bias in Language Models | Stanford HAIhai.stanford.edu
- The Dark Forest and Generative AImaggieappleton.com
- Discovering Language Model Behaviors with Model-Written Evaluations — LessWronglesswrong.com
- Projects | Karina Nguyenkarinanguyen.com
- Externalized reasoning oversight: a research direction for language model alignment — AI Alignment Forumalignmentforum.org
- Cognitive Biases in Large Language Models — LessWronglesswrong.com
- AI #24: Week of the Podcast — LessWronglesswrong.com
- A “diff” tool for AI: Finding behavioral differences in new models \ Anthropicanthropic.com
- [2203.02155] Training language models to follow instructions with human feedbackarxiv.org
- Whoever Controls Language Models Controls Politics – HANNES BAJOHRhannesbajohr.de
- How well do models follow their constitutions? — LessWronglesswrong.com