Hossein Tajdari · Pharmacist & digital health
Portfolio and writing
A view of the questions I follow in my work and writing.
About
I’m Hossein Tajdari, a pharmacist working in digital health. This page gathers the questions I return to in my work and writing.
Digital health
How technology fits into everyday health work, and where it needs to become simpler, more accessible, and more dependable.
Research and evidence
Asking what a study can actually show, where its limits are, and how carefully its findings should be used.
AI and health
Considering the usefulness of new tools alongside uncertainty, human responsibility, and the realities of use.
Selected writing
Why do I keep returning to Red Dead Redemption?Returning to an old game can be a way of noticing familiar human patterns again.Should a model sometimes say “I don’t know”?Declining to answer is not always a failure; it can be part of a sound decision.Can we check a model’s reasoning without seeing inside it?A research framework offers one approach, while human review still has real costs and limits.Health AI needs clear boundariesBefore measuring accuracy, we should define what a tool is for—and when it should step aside.Digital health should survive an internet outageOffline-first means safe saving, clear status, and understandable sync—not just an open screen.Automation does not remove errors; it moves themBefore automating a workflow, map its steps and failure points one by one.What small choices can tell us about reasoningA simple experiment can show how a system balances what it knows against the value of new information.What evidence would count for AI consciousness?Computational indicators can sharpen the debate, but they are not a definitive test of consciousness.Knowledge debt: when output outruns understandingA polished, instant result can help, but using it without understanding can create hidden costs.Can a model infer what a user knows?Training with simulated mental states may improve dialogue tasks, but it is not mind reading.How should we test an AI agent’s memory?Recalling a fact is not enough; does relevant history actually help complete a task?What can moving images reveal about GUI agents?A motion-based CAPTCHA shifts the challenge from recognizing a still image to tracking change over time.