Giovanni De Toni
Postdoctoral Fellow
ETH AI Center and ETH Zürich

Hi 👋! I am Giovanni and I am a postdoctoral fellow working with Celestine Mendler-Dünner and Simon Mayer. My research focuses on the algorithmic challenges involved in ensuring human oversight of AI systems and understanding their effects when deployed in social contexts.

I hold a PhD from the University of Trento (cum laude), advised by Bruno Lepri and Andrea Passerini. I was also part of the European Laboratory for Learning and Intelligent Systems (ELLIS) PhD network, through which I spent some time at the Max Planck Institute for Software Systems working with Manuel Gomez Rodriguez. Throughout my academic journey, I conducted research as a (visiting) scientist or intern at several institutions, including the European Commission, Google X, CERN and Fondazione Bruno Kessler. Before my PhD, I was a Research Scientist at VUI, Inc., a Boston-based startup (now acquired) developing innovative conversational AI technologies. In the past, I have also contributed to several open-source scientific libraries (e.g., Shogun).

Latest News

Publications

  1. Towards Mitigating Unwanted Recommendations with Risk-Controlling Recommender Systems
    Giovanni De Toni, Erasmo Purificato, Emilia Gomez, Andrea Passerini, Bruno Lepri, Cristian Consonni
    ACM Transactions on Recommender Systems (2026)
    [paper][code]

  2. With a Little Help From My Friends: Collective Manipulation in Risk-Controlling Recommender Systems
    Giovanni De Toni, Cristian Consonni, Erasmo Purificato, Emilia Gomez, Bruno Lepri
    FAccT: ACM Conference on Fairness, Accountability, and Transparency (2026)
    [paper][code]

  3. Multiclass Local Calibration With the Jensen-Shannon Distance
    Cesare Barbera, Lorenzo Perini, Giovanni De Toni, Andrea Passerini, Andrea Pugnana
    AISTATS (2026)
    [paper][code]

  4. Revisiting (Un)Fairness in Recourse by Minimizing Worst-Case Social Burden
    Ainhize Barrainkua, Giovanni De Toni, Jose Antonio Lozano, Novi Quadrianto
    AAAI (2026)
    Oral
    [paper][code]

  5. 🏆 You Don't Bring Me Flowers: Mitigating Unwanted Recommendations Through Conformal Risk Control
    Giovanni De Toni, Erasmo Purificato, Emilia Gomez, Andrea Passerini, Bruno Lepri, Cristian Consonni
    RecSys: 19th ACM Conference on Recommender Systems (2025)
    Best Full Paper Award at 19th ACM Conference on Recommender Systems (ACM RecSys 2025)
    [paper][code]

  6. Time Can Invalidate Algorithmic Recourse
    Giovanni De Toni, Stefano Testo, Bruno Lepri, Andrea Passerini
    FAccT: ACM Conference on Fairness, Accountability, and Transparency (2025)
    [paper][code]

  7. Towards Human-AI Complementarity with Predictions Sets
    Giovanni De Toni, Nastaran Okati, Suhas Thejaswi, Eleni Straitouri, Manuel Gomez-Rodriguez
    NeurIPS (2024)
    [paper][code]

  8. 🏆 Preference Elicitation in Interactive and User-centered Algorithmic Recourse: an Initial Exploration
    Seyedehdelaram Esfahani, Giovanni De Toni, Bruno Lepri, Andrea Passerini, Katya Tentori, Massimo Zancanaro
    ACM UMAP (2024)
    Best Short Paper Runner-up at the 32nd ACM UMAP Conference (2024)
    [paper][code]

  9. Personalized Algorithmic Recourse with Preference Elicitation
    Giovanni De Toni, Paolo Viappiani, Stefano Teso, Bruno Lepri, Andrea Passerini
    Transactions on Machine Learning Research (2024)
    [paper][code]

  10. Synthesizing explainable counterfactual policies for algorithmic recourse with program synthesis
    Giovanni De Toni, Bruno Lepri, Andrea Passerini
    Machine Learning (2023)
    [paper][code]

Preprints

  1. Divide et Calibra: Multiclass Local Calibration via Vector Quantization
    Cesare Barbera, Lorenzo Perini, Giovanni De Toni, Andrea Passerini, Andrea Pugnana
    Preprint (2026)
    [paper][code]

  2. To Ask or Not to Ask: Learning to Require Human Feedback
    Andrea Pugnana*, Giovanni De Toni*, Cesare Barbera*, Roberto Pellungrini, Bruno Lepri, Andrea Passerini
    Preprint (2025) (* equal contribution)
    [paper][code]