BERI-DMIP Collaboration
About
The BERI-DMIP Collaboration is a partnership between the Berkeley Existential Risk Initiative (BERI) and the DMIP (Data, Machines, Intelligence & People) research group at the Universitat Politecnica de Valencia (UPV) in Spain. It was established in late 2023, when BERI announced its largest cohort of new university collaborators, adding 14 groups including DMIP. BERI is a 501(c)(3) public charity that supports university research groups working to reduce existential risk by providing them with operational services and support that are difficult to obtain through normal university channels. This includes help with purchasing equipment and software, hiring research staff, covering travel and conference costs, and other administrative tasks. The DMIP group was founded in 1997 and consolidated as a group in 2000 at UPV, originally focused on extending inductive logic programming. Over time, the group broadened its research to encompass AI evaluation, intelligence measurement, machine learning, data science, predictable AI, explainable AI (XAI), AI fairness, and universal psychometrics. It is part of the VRAIN (Valencian Research Institute for Artificial Intelligence), one of Spain's largest dedicated AI research centers. The group is led by Professor Jose Hernandez-Orallo, who also serves as Director of Research at the Leverhulme Centre for the Future of Intelligence at the University of Cambridge. Other core faculty include Professors Cesar Ferri and Carlos Monserrat, and Associate Professors Maria Jose Ramirez-Quintana and Fernando Martinez-Plumed. The group has published in Nature and Science on AI evaluation topics, participated in the GPT-4 Red Team, and has worked on projects funded by DARPA (ReCOG-AI for robust AI evaluation), the Future of Life Institute (paradigms of AGI and associated risks), and the OECD (AI Futures & Skills). The collaboration received $66,000 in grant recommendations from the Survival and Flourishing Fund's 2024 S-Process round, plus a $10,000 speculation grant, with Jaan Tallinn as the recommending funder. BERI serves as the receiving charity for the collaboration's funding.
Theory of Change
The BERI-DMIP Collaboration aims to reduce existential risk from AI by supporting rigorous research on evaluating and measuring the capabilities, generality, and safety properties of AI systems. The theory of change rests on the premise that better evaluation methodologies are essential for understanding what AI systems can and cannot do, which in turn is critical for identifying risks, ensuring alignment, and informing safety policy. By providing operational support through BERI, the collaboration removes administrative barriers that slow down academic research, allowing the DMIP group to focus on developing evaluation frameworks, benchmarks, and measurement tools that can help the broader AI safety community assess and manage the risks posed by increasingly capable AI systems.
Details
- Start Date
- -
- End Date
- -
- Expected Duration
- -
- Funding Raised to Date
- $76,000
- Last Updated
- Apr 3, 2026, 1:18 AM UTC
- Created
- Apr 3, 2026, 1:18 AM UTC