20 May 2026 | Online via ZOOM
13:00 – 16:30 CEST
The workshop is co-organized by the ELLIS Unit Prague, ELLIS Associate Unit Lviv, the ELLIS Unit Alicante, and the ELLIS Program on Human-Centric Machine Learning.
.
Please register for the workshop to receive a Zoom link a few days before the workshop.
Objective:
The objective of this online workshop is to discuss recent scientific advances in the areas of
privacy, safety, trust, biases, and interpretability in artificial intelligence.
.
Organisers:
With questions please feel free to reach to:
- Rostyslav Hryniv or Tetiana Zakharchenko, ELLIS Associate Unit Lviv
- Nuria Oliver, ELLIS Unit Alicante and ELLIS Program on Human-Centric Machine Learning
- Josef Sivic, ELLIS Unit Prague
Agenda:
Each presentation slot is 30mins with a suggested 20-25 mins for presentation and 5-10 mins for questions and discussion. All times are in Central European Summer Time (CEST).
13:00-14:30 Session I
- Nuria Oliver – What foundation models learn from us and use it against us
- Mykola Pechenizkiy – When Benchmarks Mislead: Rethinking AI Safety Evaluation
- Novi Quadrianto – Performative prediction made practical
14:30-15:00 Coffee Break
15:00-16:30 Session II
- Krishna Gummadi – In Agents We Trust, but Who do Agents Trust? On latent preferences and perceptions that steer LLM generations
- Lesia Semenova – From Model Uncertainty to Trust: Safe and Interpretable AI under the Rashomon Effect
- Valentyn Melnychuk – Causal Machine Learning under Privacy Constraints (short talk)
- Denys Herasymuk – Smarter Together, Safer Apart: Achieving Privacy and Robustness in Personalized Machine Learning (short talk)
Ask your questions via SLIDO: #ellis-2026 or simply scan this QR code:
Watch the live-streaming of the event:
This event is supported by the ELIAS project – European Lighthouse of AI for Sustainability.
This project has received funding from the European Union’s Horizon Europe research and innovation programme under grant agreement 101120237.