Mel Andrews

existing work that demonstrates some of this can be found here: arxiv.org/abs/2508.02740

Who Gets Cited? Gender- and Majority-Bias in LLM-Driven Reference Selection arxiv.org

When utilized in literature review, LLMs consistently 1. fail to mention female authors in female-led literatures, 2. insist that men are more influential or more heavily cited when this is contradicted by objective citation counts, and 3. attribute women’s work to hallucinated male scholars.

The export control directive banning state of the art language models from Anthropic is only counterproductive to the US AI industry if you don’t view it as yet another act in the current production of AI safety theatrics.

Something I have noticed about attempting lit review with LLMs: Frontier models still hallucinate heavily when asked to produce citations to works in women-led literatures. When constrained to works that can be linked to, they simply re-attribute papers written by women to imaginary men.

The unprecedented leap in advancement of Anthropic’s frontier models is one of the number one global news item this week. No technical demonstration was necessary to achieve this effect. That there is successful marketing.

The irony of using an LLM-powered bot with a female-coded screen name to respond to an observation about gender bias in LLMs is not lost on me.

To the internet-savvy striving to avoid AI content: what web browsers, search engines, email servers, etc. are you using these days?

So no one catches my vibe when I say DimeNet filters (Directional Message Passing Neural Network filters for distance-wise and angular atomic positioning in molecular structure learning) look like finch chick gapes?