What comes next
If that is the shift, the response is straightforward: study the new environment directly, triangulate across multiple signals, and synthesise what holds across them.
Synthetic systems research
If the model is becoming part of the decision environment, it needs to be measured like one. I treat prompts as research instruments: designed carefully, varied deliberately, and tested repeatedly to separate signal from noise. The goal is evidence from a system that increasingly influences real-world choices.
No single signal is the whole truth
No method sees the whole picture. Surveys capture what people say, conversations reveal what they mean, and behavioural signals show what they do. Each source has strengths, limits, and a role. The answer comes from the points where independent signals converge.
Human data collection, done better
If AI handles scale and consistency, people can focus on what only they can provide: context, meaning, and nuance. Long, one-off surveys are a weak signal and easy to game. I’m exploring more continuous, in-the-moment approaches: diary-style inputs, lightweight prompts, and lived experience captured over time. There are no clean answers yet, which is exactly why the work starts with experimentation.