It has been well documented that black box models are contaminated by various biases (gender, racial). Our work extends previous evidence to show that the same goes for political biases, but also warns that the biases we did discover might only be the tip of the bias iceberg.
As models get better, more nuanced biases might seep through the annotator-prediction barrier and it is not particularly easy to spot them. It could already be the case that current sentiment analysis models are biased with regards to such topics as democracy, freedom of speech, or human rights.
Proud to share what I've been working on for the last half a year - a method to statistically assess and explain differences in the perceived meaning of concepts based on small samples.
Allow me to introduce the Supervised Semantic Differential
osf.io/preprints/ps...
For all #Severance fans out there.
Baldwin, A. L. (1942). Personal structure analysis: A statistical method for investigating the single personality. The Journal of Abnormal and Social Psychology, 37(2), 163–183. doi.org/10.1037/h006...
Current main bandwagon in AI x Psychology research has been embedding questionnaires with transformers. Are we ready to answer the question of whether transformers actually uncover the latent psychological dimensions vs. they are just useful in pinpointing any text differences?
The primary dimension of psychometric difference in LLMs is phenomenological - the extent to which they respond as if they had internal, phenomenal experience.
#AI #LLM #Psychometrics
arxiv.org/abs/2605.05080
Computational Psycholinguistics discord server is an experimental research square for people working on the cross-section of NLP and Psychology to connect, discuss ideas, learn from each other, and pick up collaborators.
You can join by following the link below
discord.gg/2DUHmcnY6Z
Get the paper at rdcu.be/eauyY
Online surveys are on their way to the grave. Are we working on alternatives? Is it time to start proctoring every survey taker?
www.nature.com/articles/d41...
