Research & Insights
Exploring how AI systems are evaluated, how organizations make decisions about them, and where existing approaches may leave important questions unanswered.
This space brings together exploratory studies, research findings, and practical observations at the intersection of AI, cognition, behavior, and organizational decision-making.
From questions to findings
Current Studies
Our studies examine how AI-related questions are handled in practice, bringing together technical, behavioral, organizational, and governance perspectives.
How Do We Really Evaluate AI?
An Exploratory Survey on AI Evaluation Practices — 2026
How do organizations evaluate AI systems in real-world settings? What evidence is used in technical assessments and due diligence? How are unexpected behaviors, governance responsibilities, and evaluation gaps addressed?
The study invites perspectives from organizations that conduct, commission, or undergo AI evaluations, including developers, deployers, research teams, investors, and procurement stakeholders.
A Research-Oriented Approach
The goal is not simply to catalogue tools or frameworks, but to understand the relationship between stated requirements, evaluation practices, and the outcomes they can — or cannot — establish.
Different roles
Including both organizations conducting evaluations and those whose systems, products, or practices are assessed.
Requirements and evidence
Distinguishing what organizations aim to assess, the methods they use, and how requirements are verified.
Beyond predefined categories
Using open-ended questions to identify difficulties, overlooked risks, and concerns not captured by standard checklists.
Research Findings
Study summaries and analyses will be published here as they become available.
Findings will be interpreted in the context of the available evidence. Where participation is limited or the sample is not representative, these limitations will be made explicit rather than presenting exploratory observations as general conclusions.
Questions or research perspectives?
GRDprocess welcomes exchanges about AI evaluation practices, methodological challenges, and research observations.