Talks & Presentations

Practical Workshop on Socio-Cultural AI Evaluation

Practical Workshop Leader · Second South American NLP School, Universidad de Buenos Aires (UBA) · Buenos Aires, Argentina · August 2026

Hands-on workshop leader at the Second South American NLP School at Universidad de Buenos Aires (UBA). Instructed graduate students and researchers on methodologies and toolkits for auditing regional stereotypes and evaluating foundation models across Latin American language varieties.


Adaptive Data Collection for Latin-American Community-sourced Evaluation of Stereotypes (LACES)

Conference Presentation & Workshop Oral · 63rd Annual Meeting of the Association for Computational Linguistics (ACL 2026) · San Diego, USA · July 2026

Presented at ACL 2026 in San Diego, California:

  • Poster presentation in the main conference Findings session on the LACES framework.
  • Oral presentation at the StereACuLT Workshop (Stereotypes Across Cultures in Language Technologies).
  • Co-organized the Birds-of-a-Feather (BoF) session “Who Gets Left Behind? Identity, Culture, and Power in NLP Systems” together with Alberto Testoni (Amsterdam UMC).

Logical Reasoning Evaluation and Social Bias

Workshop Talk · ICLR 2026 Workshop on Logical Reasoning of Large Language Models · Singapore · May 2026

Presented at the ICLR 2026 Workshop on Logical Reasoning of Large Language Models.

The presentation explored how reasoning models integrate priors from training data, how stereotypes influence step-by-step reasoning traces, and methods to detect and debias reasoning outputs.


Logical Reasoning Evaluation and Social Bias

Featured Workshop Oral & Poster · ICLR 2026 Workshop on Logical Reasoning of Large Language Models · Rio de Janeiro, Brazil · April 2026

Presented research on the intersection between formal logical reasoning in LLMs and socio-cultural bias traces. The paper was selected as one of the featured works of the workshop at ICLR 2026 in Rio de Janeiro, Brazil.


HESEIA at the India AI Impact Summit 2026

Poster & Symposium Presentation · India AI Impact Summit 2026 (AI Research & Impact Symposium) · New Delhi, India · February 2026

Represented Fundación Vía Libre at the India AI Impact Summit 2026 in New Delhi, India. Presented the HESEIA dataset (46,499 sentences co-designed in real school settings across Latin America) within the AI Research & Impact Symposium.


Insights from a Disaggregated Analysis of Kinds of Biases in a Multicultural Dataset

Workshop Presentation · 9th Widening NLP Workshop (WiNLP 2025) · Online / Suzhou · November 2025

Presented at the 9th Widening NLP Workshop (WiNLP 2025).

The talk demonstrated how aggregate bias metrics mask disparities across individual demographics and presented qualitative and quantitative insights from evaluating multicultural datasets.


HESEIA: A Community-Based Dataset for Evaluating Social Biases in Large Language Models

Oral & Poster Presentation · Conference on Empirical Methods in Natural Language Processing (EMNLP 2025) · Suzhou, China · November 2025

Oral presentation and poster at EMNLP 2025 presenting HESEIA, a community-based dataset for evaluating social biases in large language models co-designed with 370 teachers and 5,000+ students across 189 schools in Latin America.


Navigating Ethical Challenges in NLP: Hands-on Strategies for Students and Researchers

Tutorial · 63rd Annual Meeting of the Association for Computational Linguistics (ACL 2025) · Vienna, Austria · July 2025

Tutorial presented at ACL 2025 in Vienna, Austria, together with Luciana Benotti, Fanny Ducel, Karën Fort, Zhijing Jin, Min-Yen Kan, Seung-Hun J. Lee, Minzhi Li, Margot Mieskes, and Adriana Pagano.

The tutorial provides concrete methodology and interactive exercises for identifying, characterizing, and mitigating ethical harms in NLP research, from data collection and annotation to model auditing and deployment.


Socio-Cultural Evaluation of AI

Practical Session & Tutorial · Khipu: Latin American Meeting in Artificial Intelligence · Latin America · March 2025

Practical session presented at Khipu (Latin American Meeting in AI).

The tutorial covered practical tools and datasets (including SeeGULL and HESEIA) to systematically test generative AI models for regional stereotypes, cultural grounding, and demographic harms across Latin American varieties.


Detecting Correct Answers to Open Questions and Its Impact on Language Models' Confidence Scores

Oral Presentation · LatinX in AI (LXAI) Research Workshop at NAACL 2024 · Mexico City, Mexico · June 2024

Presented at the LatinX in AI (LXAI) Research Workshop colocated with NAACL 2024 in Mexico City, Mexico.

This presentation discussed calibration challenges in LLMs when answering open-ended questions, focusing on the effect of automated evaluation metrics on measured confidence scores and calibration error.


EDIA: Evaluation of Bias for Experts in Discrimination, Not in Computer Science

Tutorial & CRAFT Session · ACM Conference on Fairness, Accountability, and Transparency (ACM FAccT 2024) · Rio de Janeiro, Brazil · June 2024

Delivered an interactive tutorial and demonstration of EDIA (Evaluation of Discrimination in AI) at ACM FAccT 2024 in Rio de Janeiro, Brazil, together with Laura Alonso Alemany. Also participated in the CRAFT session ‘Towards Responsible Non-extractive AI Research’.


Workshop on Responsible AI & Bias Evaluation

Workshop Organizer & Speaker · FaMAF, Universidad Nacional de Córdoba · Córdoba, Argentina · May 2024

Main organizer and instructor for the Responsible AI Workshop held at FaMAF, Universidad Nacional de Córdoba. Funded through a KhipuX / Americas Fund grant.


Evaluating Bias in Language Models with the EDIA Tool

Workshop Speaker & Tool Demo · RightsCon 2023 · San José, Costa Rica · June 2023

Presented the open-source EDIA tool at RightsCon 2023 in San José, Costa Rica, showcasing methods for civil society and domain experts to evaluate biases in language models without requiring machine learning expertise.