The Institute of Science and Technology Austria (ISTA) launched its Responsible AI Program today, backed by a €5 million donation from Canadian entrepreneur and Uber co-founder Garrett Camp. The first four postdoctoral researchers have been selected to advance work on trustworthy, ethical AI, focusing on privacy, transparency, reliability, and safety.
The program funds PhD students, postdocs, scientific interns, and residents whose projects promote fairness, transparency, privacy, and accountability. A competitive evaluation by ISTA's Responsible AI Committee-made up of experts from inside and outside the institute-selected the inaugural cohort of postdocs from a strong pool of early-career applicants.
Projects targeting privacy, safety, and efficiency
Joel Daniel Andersson (Monika Henzinger group) will develop memory-efficient, mathematically sound methods for differential privacy. His work aims to let AI models learn from sensitive data without exposing it, so models remain accurate and useful.
Filip Cano (Thomas Henzinger group) will design runtime interventions that enforce safety and fairness in AI decision-makers as conditions change. The goal is to create explainable, flexible corrections that build trust without unnecessary disruption.
Valentino Maiorca (Francesco Locatello group) will investigate how to identify which internal modules of large transformer models are needed for a given input. Skipping unnecessary computation could make AI systems more interpretable, flag unreliable predictions, and reduce energy consumption.
Marco Pegoraro (Alexander Bronstein group) will focus on improving confidence estimates in AI models for computational biology, particularly protein structure prediction. His project examines why confidence scores can be misleading and develops methods to make them more reliable, helping researchers decide which predictions to trust.
Long-term commitment to responsible AI
ISTA President Martin Hetzer said: "The Responsible AI Program reflects ISTA's vision of advancing AI research that is both scientifically rigorous and socially meaningful. Thanks to Garrett Camp's generous support, we are empowering young scientists to explore groundbreaking questions about how AI can be designed, understood, and deployed responsibly."
The program will run three annual calls, giving early-career researchers at ISTA ongoing opportunities to contribute. The donation, announced in January, was Camp's first philanthropic gift to a European research institute.
Why this matters for science and research
Each project tackles a practical barrier that researchers face when deploying AI in labs and real-world settings-privacy in streaming data, safe real-time decisions, efficient model inference, and trustworthy confidence scores. As AI becomes central to fields like biology and physics, understanding these challenges is essential. For professionals looking to deepen their knowledge of AI's role in these domains, AI for Science & Research offers training and resources tailored to the needs of working scientists.
Your membership also unlocks:
