copenlu/cmt-benchmark-counterfact
Viewer
•
Updated
•
1.8k
•
47
CopeNLU
university
AI & ML interests
natural language understanding, low-resource learning, fact checking, explainable AI
Recent Activity
View all activity
copenlu's activity
LoÂ
updated
a
dataset
2 days ago
LoÂ
published
a
dataset
2 days ago
KarolinaÂ
authored
2
papers
7 days ago
IAugensteinÂ
authored
a
paper
16 days ago
NadavÂ
authored
a
paper
16 days ago
zainmujahidÂ
authored
a
paper
17 days ago
IAugensteinÂ
authored
a
paper
20 days ago
zainmujahidÂ
authored
a
paper
20 days ago
gretawarrenÂ
authored
a
paper
20 days ago
dwright37Â
authored
a
paper
20 days ago
frimelleÂ
authored
a
paper
21 days ago
zainmujahidÂ
authored
a
paper
21 days ago
frimelleÂ
authored
a
paper
21 days ago
Post
2381
What’s in a name? More than you might think, especially for AI.
Whenever I introduce myself, people often start speaking French to me, even though my French is très basic. It turns out that AI systems do something similar:
Large language models infer cultural identity from names, shaping their responses based on presumed backgrounds. But is this helpful personalization or a reinforcement of stereotypes?
In our latest paper, we explored this question by testing DeepSeek, Llama, Aya, Mistral-Nemo, and GPT-4o-mini on how they associate names with cultural identities. We analysed 900 names from 30 cultures and found strong assumptions baked into AI responses: some cultures were overrepresented, while others barely registered.
For example, a name like "Jun" often triggered Japan-related responses, while "Carlos" was linked primarily to Mexico, even though these names exist in multiple countries. Meanwhile, names from places like Ireland led to more generic answers, suggesting weaker associations in the training data.
This has real implications for AI fairness: How should AI systems personalize without stereotyping? Should they adapt at all based on a name?
Work with some of my favourite researchers: @sidicity Arnav Arora and @IAugenstein
Read the full paper here: Presumed Cultural Identity: How Names Shape LLM Responses (2502.11995)
Whenever I introduce myself, people often start speaking French to me, even though my French is très basic. It turns out that AI systems do something similar:
Large language models infer cultural identity from names, shaping their responses based on presumed backgrounds. But is this helpful personalization or a reinforcement of stereotypes?
In our latest paper, we explored this question by testing DeepSeek, Llama, Aya, Mistral-Nemo, and GPT-4o-mini on how they associate names with cultural identities. We analysed 900 names from 30 cultures and found strong assumptions baked into AI responses: some cultures were overrepresented, while others barely registered.
For example, a name like "Jun" often triggered Japan-related responses, while "Carlos" was linked primarily to Mexico, even though these names exist in multiple countries. Meanwhile, names from places like Ireland led to more generic answers, suggesting weaker associations in the training data.
This has real implications for AI fairness: How should AI systems personalize without stereotyping? Should they adapt at all based on a name?
Work with some of my favourite researchers: @sidicity Arnav Arora and @IAugenstein
Read the full paper here: Presumed Cultural Identity: How Names Shape LLM Responses (2502.11995)
IAugensteinÂ
authored
5
papers
21 days ago
SemEval 2017 Task 10: ScienceIE - Extracting Keyphrases and Relations from Scientific Publications
Paper
•
1704.02853
•
Published
•
1
A Supervised Approach to Extractive Summarisation of Scientific Papers
Paper
•
1706.03946
•
Published
•
1
MultiFC: A Real-World Multi-Domain Dataset for Evidence-Based Fact Checking of Claims
Paper
•
1909.03242
•
Published
•
1
Why Should This Article Be Deleted? Transparent Stance Detection in Multilingual Wikipedia Editor Discussions
Paper
•
2310.05779
•
Published
•
1
SubjQA: A Dataset for Subjectivity and Review Comprehension
Paper
•
2004.14283
•
Published
•
1