The oversight problem nobody is measuring

Diagnostic AI is getting better quickly. Radiology tools, sepsis alerts, triage support systems. Clinicians are using them daily, and in many cases they are catching things that would otherwise be missed.

The problem is what happens in the other direction. When a clinician reviews an AI output rather than forming an independent assessment first, the habit of independent assessment gradually weakens. Not dramatically. Not all at once. Just quietly, over months and years.

Both FDA guidance and NHS governance frameworks assume a clinician who can genuinely override an AI recommendation when the clinical picture warrants it. That assumption deserves scrutiny. Oversight is not the same as review.

What most trusts are training for, and what they are not

Most clinical AI training covers how to use the tools correctly. Workflow integration, data inputs, understanding confidence scores, knowing when to escalate. That is necessary work.

What it does not cover is the cognitive pattern underneath. Specifically, how repeated reliance on algorithmic prompts changes the way clinicians read a case before the system has flagged anything. The instinct for what looks wrong before the algorithm says so is a trained capacity. It erodes with disuse like any other.

That gap is not visible in audit data, satisfaction scores, or compliance checklists. It tends to become visible only in retrospect, when something gets missed that an experienced clinician would once have caught.

What Steve covers with clinical audiences

Steve speaks to medical directors, clinical leads, and L&D teams about the specific ways AI tool use reshapes clinical reasoning over time. The talk is grounded in cognitive science and draws on documented patterns in how professional judgement degrades when it is regularly outsourced.

The session is not a warning against using AI tools. It is an argument for deliberately maintaining the diagnostic instincts that make human oversight real rather than nominal. That includes what organizations can do structurally, not just what individual clinicians should practice.

Steve also speaks at clinical conferences. Talks can be adapted for general clinical audiences or focused on a specific discipline, department, or governance context.

Topics for Healthcare audiences

Steve speaks to healthcare organizations on the following topics. Each can be delivered as a keynote, half-day workshop, or executive briefing.

Who books Steve

Medical directors, clinical leads, L&D teams at NHS trusts and hospital networks, conference organisers for clinical conferences.

If you are planning an event and want to discuss whether Steve's work is a good fit, the fastest route is a short conversation. No pitch deck required.