writing
Explain the boundary, not just the method
Two public articles on equity and responsible AI in public health.
- Role
- Author
- Status
- Two published explanatory articles
Context and real problem
Public-health AI can be difficult to discuss without either hiding the method or overstating what it can tell us. An explanation needs to make the analytical choices, uncertainty, and human responsibilities understandable.
Andres’s role
I wrote two Phiz articles on equity in AI for public health. The first explains why safeguards should shape the work from the outset; the second follows how those safeguards affect an exploratory youth-persona workflow.
Collaborators and setting
The articles draw on the public account of work at WDGPH through the AI4PH internship. They explain an analytical setting and its boundaries, rather than claiming to speak for every institution or contributor.
Methods and tools
The writing connects survey design, missingness, clustering, interpretability, and human review. It also explains NIST AI RMF 1.0 through concrete decisions about governing, mapping, measuring, and managing risk.
Responsibility and safety
The articles contain no sensitive findings, subgroup results, cluster sizes, or draft persona labels. Plain language should preserve the limits of the evidence, including the option to restrict an output or decide not to create a persona.
Outputs and outcomes
Part 1 was published on August 5, 2026; Part 2 on August 19, 2026. Both carry my named byline. These are public explanatory articles, distinct from peer-reviewed research papers.
What I learned
A good explanation makes room for a reader to question the work. Clarity includes saying what a method cannot establish, what still needs judgment, and which uses should remain off limits.
Current status and links
Both articles are publicly available on Phiz. Their links provide the full account; this site presents a short summary without republishing internal work or changing the articles’ publication category.