Large Language Models Are Biased Because They Are Large Language Models
2025/01/01 by Philip Resnik · 4 voices · 1 citation
Computer Science · #Topic Modeling
paper · doi:10.1162/coli_a_00558
Abstract
Abstract This position paper’s primary goal is to provoke thoughtful discussion about the relationship between bias and fundamental properties of large language models (LLMs). I do this by seeking to convince the reader that harmful biases are an inevitable consequence arising from the design of any large language model as LLMs are currently formulated. To the extent that this is true, it suggests that the problem of harmful bias cannot be properly addressed without a serious reconsideration of AI driven by LLMs, going back to the foundational assumptions underlying their design.
Citations
Cited by
Discussions
- Is there a relationship between bias and fundamental properties of LLMs? While LLMs are powerful, they contain harmful biases that can emerge unpredictably in their behavior. Read more on this in this [bsky, 17 points, 0 comments]
- Philip Resnik: ‘to address LLM bias, start from the hard-core interpretation of the distributional hypothesis, where meaning is distribution, as opposed to distribution being part of what goes into th [bsky, 10 points, 0 comments]
- doi.org/10.1162/coli... [bsky, 2 points, 0 comments]
- An excellent (and provoking) position paper from Philip Resnik in the latest issue of Computational Linguistics! doi.org/10.1162/coli... [bsky, 0 points, 0 comments]
Related