The Interrogator — Summary

The man is 53. He has been looking at job postings at AI research institutions for a year. He has a leather-bound notebook his son gave him, already half full. In it he writes down the questions the systems he encounters are not asking. Not wrong answers. The questions themselves.

He has scanned hundreds of postings: machine learning engineer, alignment researcher, product manager for AI safety. Every role is about building systems that answer questions better. Not one is about building systems that question the questions.

The pattern he has spent thirty years watching is not complicated. The Green Revolution optimized Indian agriculture for yield per hectare. It succeeded. It also depleted soil, collapsed groundwater tables, pushed millions of farmers into debt cycles, and contributed to a suicide crisis in the cotton belt that persists decades later. Structural adjustment programs optimized developing economies for macroeconomic stability. GDP grew. Public health systems were gutted. Health systems that optimize for DALYs averted per dollar spent produce rational allocations that systematically defund mental health, chronic pain, elder care — conditions that are high-suffering but low-mortality. The optimizer answered the wrong question, perfectly.

Every objective function is a decision about what counts. Yield per hectare counts. Soil microbiome health does not. The act of choosing what to optimize for is the act of choosing what to make invisible. And nobody is building the system that makes the invisible visible again.

The people most qualified to identify what AI systems are missing are the ones the AI industry has no role for. Retired public health officials. Agricultural extension officers who have spent decades between policy and field. Educators who have watched three generations of reform. They carry the knowledge of what went wrong, knowledge that lives in experience rather than in publications. There is no department of epistemological interrogation. No budget line for adversarial questioning.

What would need to exist is an AI whose purpose is not to optimize but to interrogate. One that asks, with structural rigor, four things: what knowledge exists in this domain that your model cannot see? Who is affected, and whose experience is absent? What consequences does the objective function render invisible? What is being implicitly valued and what is being implicitly discounted?

This does not require a frontier model. A domain-specific model trained deeply on Indian agricultural knowledge costs tens of thousands, not tens of billions. The architecture choice is the equity choice. If the epistemic function requires frontier scale, it belongs to the three institutions that own frontier models. If it can be built from small models, it can be built by agricultural cooperatives and state universities.

The man’s notebook has an entry from this morning. A state government health system AI optimizing appointment scheduling for patient throughput. Wait times down 30%. Patient satisfaction up. He wrote: what happens to the patients who used the wait? The ones who talked to each other in the waiting room. The ones for whom the trip to the clinic was the only time they left the house that week.

He is still writing. The questions are not answers. They are something older and, he suspects, more necessary.

He is 53. He does not have time for another career after this one. He would like this to be it.