Research Internships in Credible AI
A summer research program at the Centre for Credible AI in Warsaw and Jagiellońskie Centrum Sztucznej Inteligencji in Krakow.
RICA Fellowships offer students the opportunity to spend two months of the summer working on a research project in the field of Credible AI, under the supervision of an experienced mentor. The program is designed to give participants hands-on research experience, expose them to current problems in the field, and help them develop the skills needed for a career in AI research.
Fellows join the Centre for Credible AI in Warsaw or Jagiellońskie Centrum Sztucznej Inteligencji in Krakow, where they become part of an active research community and contribute to ongoing work on making AI systems more explainable, reliable, and aligned with human values and objectives.
The program is open (but not restricted) to students, regardless of their year of study or field of specialization. We welcome applications from people with backgrounds in computer science, mathematics, statistics, cognitive science, philosophy, and related disciplines.
Strong candidates typically have:
- A genuine interest in AI research and questions of AI credibility, safety, or alignment
- Solid technical foundations relevant to their proposed project
- The ability to work independently while collaborating effectively with a mentor
- Good written and spoken English
- A two-month research project carried out during the summer or fall. The exact start and end dates are agreed individually with the mentor to accommodate academic calendars and project needs.
- One-on-one mentorship from a researcher working on Credible AI throughout the duration of the project.
- A stipend covering the cost of living in Warsaw or Krakow for the duration of the fellowship, so that fellows can focus fully on their research (~5k PLN).
- Access to the Centre's research environment, including seminars, reading groups, and informal discussions with other researchers and fellows.
Projects are developed jointly with mentors and span a variety of topics in the area of Credible AI. A list of mentors and example project ideas will be published ahead of each application cycle.
- Automatic Autism Spectrum Disorder detection based on amateur videos with children, more...
- From Actions to Meaning: Do AI Models Build World Models? more...
- Reliable Multi-Agent Reasoning with Verifiable Rewards, more...
- Stress-testing AI solutions in LDCT lung cancer diagnostics, more...
- Training an RLHF agent to play Ortho at the Copernicus Science Centre, more...
- Building Modelpedia: infrastructure for a shared repository of model findings, more...
- Unlearning and Unbranding in Generative AI: Testing the Limits of Concept Erasure more...
- Object-Centric Learning with Language Explanations more...
- Sparse Neural Networks with Learnable Structured Connectivity via Laplace Kernel Operators more...
- Interpretable Graph Neural Networks for Drug-Target Binding Prediction more...
- Enforcing Monosemanticity in Vision Models via Counterfactual Generative Regularization more...
Applications for the programme can be submitted by 30 May via the google form
We will contact the selected candidates by 7 June
If you have any questions about the RICA programme, please report them as an issue in this GitHub repository: https://github.com/CredibleAI/RICA/issues
Centre for Credible AI · Warsaw, Poland