Lorenzo Lupo

I am an interdisciplinary researcher with expertise in machine learning and natural language processing applied to the social sciences, with a particular interest in the societal impacts of AI.
My recent work includes evaluating the influence of AI on knowledge platforms such as Wikipedia, investigating LLMs' simulation capabilities of entire demographic groups, and analyzing legal and social media texts to uncover historical trends and shifts in national cultures.
After completing my PhD on neural machine translation at Grenoble Alpes University, I held research positions at the University of Milan and Bocconi University, where I currently work as a freelance researcher, collaborating with Kai Zhu, Nicolò Cavalli, the MilanNLP research group, and many other researchers.

Publications

SimBench: Benchmarking the Ability of Large Language Models to Simulate Human Behaviors
Tiancheng Hu, Joachim Baumann, Lorenzo Lupo, Dirk Hovy, Nigel Collier, Paul Röttger
under review at NeurIps

Male agency? Analyzing fatherhood roles in Swedish Parliamentary documents 1993-2021
Lorenzo Lupo, Oscar Magnusson, Elin Naurin, Dirk Hovy, Lena Wangnerud
under review at Politics & Gender

Beyond the Stats: Realities, Perception, and Social Media Discourse on Poverty
Paul Bose, Lorenzo Lupo, Mahyar Habibi, Dirk Hovy, Carlo Schwarz
AEA Papers and Proceedings 2024

DADIT: A Dataset for Demographic Classification of Italian Twitter Users and a Comparison of Prediction Methods
Lorenzo Lupo, Paul Bose, Mahyar Habibi, Dirk Hovy, Carlo Schwarz
LREC-COLING 2024

How to Use LLMs for Text Coding: The Case of Fatherhood Roles in Public Policy Documents
Lorenzo Lupo, Oscar Magnusson, Dirk Hovy, Elin Naurin, Lena Wängnerud
preprint on arXiv

Encoding Sentence Position in Context-Aware Neural Machine Translation with Concatenation
Lorenzo Lupo, Marco Dinarelli, Laurent Besacier
Insights 2023

Focused Concatenation for Context-Aware Neural Machine Translation
Lorenzo Lupo, Marco Dinarelli, Laurent Besacier
WMT 2022

Divide and Rule: Effective Pre-Training for Context-Aware Multi-Encoder Translation Models
Lorenzo Lupo, Marco Dinarelli, Laurent Besacier
ACL 2022

Optimistic Policy Optimization via Multiple Importance Sampling
Matteo Papini, Alberto Metelli, Lorenzo Lupo, Marcello Restelli
ICML 2019

Tutoring

• 11/2023 - current: Alex Lucchini on Mapping the EU Political Landscape on Population Ageing.
• 09/2023 - current: Michele Scarnera on Job Shop Scheduling with Priority Function Embeddings.
• 09/2023 - current: Derya Durmush on the MENTALISM project.
• 02/2021 - 08/2021: Laura Alonzo Canul on Efficient Transformers for long context-aware neural machine translation.

Teaching

• 01/2024 - 03/2024: Computational Methods for Political and Social Sciences
  @ Bocconi University (PhD course)
• 01/2022 - 04/2022: Natural language processing
  @ Polytech Grenoble - Université Grenoble Alpes | assignments (excercice classes, M1)
• 01/2022 - 04/2022: Natural language processing for finance
  @ ENSIMAG - Université Grenoble Alpes | assignments (excercice classes, M2)
• 01/2021 - 04/2021: Natural language processing
  @ Polytech Grenoble - Université Grenoble Alpes | assignments (excercice classes, M1)
• 01/2021 - 04/2021: Natural language processing for finance
  @ ENSIMAG - Université Grenoble Alpes | assignments (excercice classes, M2)

Communications

• 01/07/2025: "Benchmarking the Ability of Large Language Models to Simulate Humans"
  @ Summer Institutes of Computational Social Science 2025 , Como, Italy. Slides here.
• 27/05/2025: "Social Media Text as Data"
  @ University of Milan's Department of Social and Political Sciences , Milan, Italy. Slides here.
• 23/05/2024: "DADIT: A Dataset for Demographic Classification of Italian Twitter Users and a Comparison of Prediction Methods"
  @ LREC-COLING 2024 , Turin, Italy. Poster here.
• 21-22/04/2024: MENTALISM Conference
  Co-organized the MENTALISM Conference on Computational Social Sciences @Bocconi University.
• 18/03/2024: "Towards Human-Level Text Coding with LLMs"
  @ DAISI Breakfast Bytes , Milan, Italy. Slides here.
• 26/10/2023: "Value Kaleidoscope: Engaging AI with Pluralistic Values" by Sorensen et al.
  @ MilanNLP reading group, Milan, Italy. Slides here.
• 26/10/2023: "The uselessness of AI ethics" by Luke Munn
  @ MilanNLP reading group, Milan, Italy. Slides here.
• 11/05/2023: "One Embedder, Any Task: Instruction-Finetuned Text Embeddings" by Su et al.
  @ MilanNLP reading group, Milan, Italy. Slides here.
• 28/03/2023: Challenges and Remedies for Context-Aware Neural Machine Translation
  @ PhD defence, Grenoble, France. Slides here.
• 07/12/2022: Focused concatenation for context-aware neural machine translation
  @ 7th Conference on Machine Translation, Abu Dhabi, UAE. Slides here.
• 05/12/2022: Towards document-level neural machine translation
  @ MIAI-DAYS 2022, Grenoble, France.
• 23/06/2022: Divide&Rule: pre-training for context-aware multi-encoder translation models
  @ 60th Annual Meeting of the ACL, Dublin, Ireland.
• 05/10/2021: Divide&Rule: pre-training for context-aware multi-encoder translation models
  @ Naver Labs Europe, Grenoble, France.
• 10/07/2021: Recent advances in document-level neural machine translation
  @ LIG-GETALP, Grenoble, France.

Courses

• 04-06/2024: AI Governance
  Attended an in-depth course about AI Governance by AI Safety Fundamentals.
• 01-04/2024: BEAT Program on Advanced Teaching
  Attended a course on advanced teaching organized by Bocconi University and targeting Professors and teaching postdocs (like myself).