Analysis and commentary on AI validation, evidentiary tradecraft, and defensible AI-augmented research.
Variable-ratio reward and sycophancy, examined as one mechanism running through AI companionship, workplace AI trust, and the limits of AI reliability alike - with a look at where this road has already been travelled once before, by tobacco.
Read Case FileFive genuinely unresolved questions raised by human-AI reasoning divergence, explained in plain language, with the case for each possible answer weighed honestly rather than forced toward a tidy conclusion.
Read Case FileOne failure pattern, recurring at three scales - individual reasoning, live AI correspondence, and institutional governance. What an anticipatory alternative would actually require.
Read Case FileNIST, ISO 42001, and the EU AI Act were finalised before agentic AI deployment scaled - and still contain no reference to it. A documented, measurable gap, examined against a live case in South Australia.
Read Case FileA case study in human-AI reasoning divergence, and what five decades of cognitive-science research already say about why careful thinking alone doesn't guarantee accuracy.
Read Case FileWhy evidence architecture matters more than content moderation - and where DPI fits in a governance conversation dominated by disclosure rules and infrastructure control.
Read Case FileThis section will also carry longer-form written analysis. Check back soon, or get in touch if you'd like to be notified when new pieces go live.