Lucas Monteiro Paes
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Lucas Monteiro Paes has a total of 9 co-authors. Here are the top co-authors by number of shared publications:
- Flavio du Pin Calmon
John A. Paulson School of Engineering and Applied Sciences, Harvard University, Cambridge, MA, USA
3 shared publications - Caio Cesar Vieira Machado
University of Oxford, Oxford, United Kingdom; Harvard University, Cambridge, USA and University of Sao Paulo, São Paulo, Brazil
2 shared publications Ahmad Beirami
Google Research, New York, NY, USA
1 shared publicationAlex Beutel
OpenAI, San Francisco, CA, USA
1 shared publicationAnanda Theertha Suresh
Google Research, New York, NY, USA
1 shared publication- Claudio Mayrink Verdun
Harvard University, Cambridge, USA
1 shared publication
Latest Publications
- Open Access
- 23 June 2025
- Published by Association for Computing Machinery (ACM)
- p. 3046-3074
AI Alignment at Your Discretion
- 23 June 2025
- Published by Association for Computing Machinery (ACM)
- p. 3046-3074
Open AccessAI Alignment at Your Discretion
- Open Access
- 3 June 2024
- Published by Association for Computing Machinery (ACM)
- p. 2234-2253
Algorithmic Arbitrariness in Content Moderation
Machine learning (ML) is widely used to moderate online content. Despite its scalability relative to human moderation, the use of ML introduces unique challenges to content moderation. One such challenge is predictive multiplicity: multiple competing models for content classification may perform equally well on average, yet assign conflicting predictions to the same content. This multiplicity can result from seemingly innocuous choices made during training, which do not meaningfully change the accuracy of the ML model, but can nevertheless change what the model gets wrong. We experimentally demonstrate how content moderation tools can arbitrarily classify samples as “toxic,” leading to arbitrary restrictions on speech. We use the principles set by the International Covenant on Civil and Political Rights (ICCPR), namely freedom of expression, non-discrimination, and procedural justice to interpret the effects of these findings in terms of Human Rights. We analyze (i) the extent of predictive multiplicity among popular state-of-the-art LLMs used for detecting “toxic” content; (ii) the disparate impact of this arbitrariness across social groups; and (iii) the magnitude of model multiplicity on content that is unanimously recognized as toxic by human annotators. Our findings indicate that the up-scaled algorithmic moderation risks legitimizing an “algorithmic leviathan”, where an algorithm disproportionately manages human rights. To mitigate such risks, our study underscores the need to identify and increase the transparency of arbitrariness in content moderation applications. Our findings have implications to content moderation and intermediary liability laws being discussed and passed in many countries, such as the Digital Services Act in the European Union, the Online Safety Act in the United Kingdom, and the recent TSE resolutions in Brazil.
- 3 June 2024
- Published by Association for Computing Machinery (ACM)
- p. 2234-2253
Open AccessAlgorithmic Arbitrariness in Content Moderation
Machine learning (ML) is widely used to moderate online content. Despite its scalability relative to human moderation, the use of ML introduces unique challenges to content moderation. One such challenge is predictive multiplicity: multiple competing models for content classification may perform equally well on average, yet assign conflicting predictions to the same content. This multiplicity can result from seemingly innocuous choices made during training, which do not meaningfully change the accuracy of the ML model, but can nevertheless change what the model gets wrong. We experimentally demonstrate how content moderation tools can arbitrarily classify samples as “toxic,” leading to arbitrary restrictions on speech. We use the principles set by the International Covenant on Civil and Political Rights (ICCPR), namely freedom of expression, non-discrimination, and procedural justice to interpret the effects of these findings in terms of Human Rights. We analyze (i) the extent of predictive multiplicity among popular state-of-the-art LLMs used for detecting “toxic” content; (ii) the disparate impact of this arbitrariness across social groups; and (iii) the magnitude of model multiplicity on content that is unanimously recognized as toxic by human annotators. Our findings indicate that the up-scaled algorithmic moderation risks legitimizing an “algorithmic leviathan”, where an algorithm disproportionately manages human rights. To mitigate such risks, our study underscores the need to identify and increase the transparency of arbitrariness in content moderation applications. Our findings have implications to content moderation and intermediary liability laws being discussed and passed in many countries, such as the Digital Services Act in the European Union, the Online Safety Act in the United Kingdom, and the recent TSE resolutions in Brazil.
- 9 May 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal on Selected Areas in Information Theory
- Vol. 5, 659-674
Multi-Group Fairness Evaluation via Conditional Value-at-Risk Testing
Machine learning (ML) models used in prediction and classification tasks may display performance disparities across population groups determined by sensitive attributes (e.g., race, sex, age). We consider the problem of evaluating the performance of a fixed ML model across population groups defined by multiple sensitive attributes (e.g., race and sex and age). Here, the sample complexity for estimating the worst-case performance gap across groups (e.g., the largest difference in error rates) increases exponentially with the number of group-denoting sensitive attributes. To address this issue, we propose an approach to test for performance disparities based on Conditional Value-at-Risk (CVaR). By allowing a small probabilistic slack on the groups over which a model has approximately equal performance, we show that the sample complexity required for discovering performance violations is reduced exponentially to be at most upper bounded by the square root of the number of groups. As a byproduct of our analysis, when the groups are weighted by a specific prior distribution, we show that Rényi entropy of order 2/3 of the prior distribution captures the sample complexity of the proposed CVaR test algorithm. Finally, we also show that there exists a non-i.i.d. data collection strategy that results in a sample complexity independent of the number of groups.
- 9 May 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal on Selected Areas in Information Theory
- Vol. 5, 659-674
Multi-Group Fairness Evaluation via Conditional Value-at-Risk Testing
Machine learning (ML) models used in prediction and classification tasks may display performance disparities across population groups determined by sensitive attributes (e.g., race, sex, age). We consider the problem of evaluating the performance of a fixed ML model across population groups defined by multiple sensitive attributes (e.g., race and sex and age). Here, the sample complexity for estimating the worst-case performance gap across groups (e.g., the largest difference in error rates) increases exponentially with the number of group-denoting sensitive attributes. To address this issue, we propose an approach to test for performance disparities based on Conditional Value-at-Risk (CVaR). By allowing a small probabilistic slack on the groups over which a model has approximately equal performance, we show that the sample complexity required for discovering performance violations is reduced exponentially to be at most upper bounded by the square root of the number of groups. As a byproduct of our analysis, when the groups are weighted by a specific prior distribution, we show that Rényi entropy of order 2/3 of the prior distribution captures the sample complexity of the proposed CVaR test algorithm. Finally, we also show that there exists a non-i.i.d. data collection strategy that results in a sample complexity independent of the number of groups.