01 / Necessary informational conditions
From notice to
contestable grounds.
A person can receive an explanation and still lack the information needed to test it.
My work on necessary informational conditions asks what must be available when a decision is already required, or promised, to be contestable. The starting point is a specific concern: a questionable record, an unsuitable standard, an unsupported inference, or a mismatch between a general rule and an individual case.
The framework distinguishes basic orientation, the preservation of the grounds of a decision, and access activated by a concrete inquiry. A proposed information requirement has to earn its place: if a feasible substitute supports the same assessment task, the original requirement is not necessary in that setting.
This is a developing framework, with bounded claims and conditions for revision. The research program includes conceptual papers and a proposed empirical study.
02 / Human–AI decision-making
Independent evidence
should have a way in.
My proposed empirical work asks how the order of evidence and advice—and a person’s prior commitment to a judgment—affect the weight given to valid independent evidence. Matched AI and human advisers would help distinguish the influence of an AI label from the influence of advice itself.
A connected study would use synthetic consumer-credit decisions to examine a narrower question: which information helps people identify a particular defect, and when can another form of information do the same work?
The aim is to measure the quality of inquiry, including mistaken objections, effort, and conditions in which additional information does not help. These studies are proposed; data collection and results are still ahead.
03 / Institutions and responsibility
What makes a judgment
answerable?
A second line of conceptual work asks when private organizations have a responsibility to respond to challenges to their judgments. It examines consequential effects, the control of relevant evidence, and the availability of other meaningful routes for inquiry.
The question of whether answerability is owed is distinct from the question of what information a meaningful challenge requires. Keeping them separate makes both arguments more precise.
04 / A developing research practice
Make the question precise.
Leave room to be wrong.
My approach moves between conceptual analysis, institutional cases, and empirical questions. Practical experience helps identify the problem; scholarship helps define it; carefully designed studies can test the parts that call for measurement.
I am strengthening my preparation in experimental design, statistical modeling, and measurement. Planned work includes preregistration, documented materials, reproducible analyses, and careful reporting of null results, burdens, and uneven effects.
Across this work, I want to advance knowledge about when reliance on AI is warranted and what makes responsible criticism possible.
05 / Beyond the paper
Knowledge people
can work with.
My longer-term goal is to connect academic research with the people who use, design, and encounter consequential decision systems.
- Public understanding. Develop clear explanations and worked examples that preserve the distinctions people need to evaluate a claim.
- Reusable research infrastructure. Prepare accessible synthetic materials and documented tools that other researchers and educators can examine and adapt.
- More routes into inquiry. Create opportunities for students with practical experience to learn research design, responsible data practice, and interpretation.
- Collaboration across settings. Build conversations among researchers, civic and nonprofit organizations, and people with direct experience of consequential decisions.
These are commitments guiding the work ahead. DG Humana Institute provides a developing home for its broader public purpose.
Visit the research project