Clinicians Want AI Collaboration and Teaching Workflow Skills

Clinicians want training they can use inside the encounter, the operating room, and the feedback conversation. Examples from oncology, primary care, and AUA25 urology education show that CME limited to awareness is missing the operational work clinicians need to do.

AI education needs to move into the clinical encounter

The AI conversation this week was less about whether clinicians should use AI and more about how they should work with it without surrendering judgment.

One clinician-educator discussion framed ambient scribes as useful because they can get technology out of the way: less note-taking, more eye contact, more attention to the patient story. The same conversation also raised the harder issue for education teams: clinicians need to know how to prompt models with the right context, judge whether outputs fit a local population, and recognize when bias or workflow barriers make the tool less useful than it appears (AI and Healthcare).

That caution showed up in oncology. A randomized trial summary shared by Fumiko Ladd Chino described AI-triggered emails to oncologists about genomically matched trials that did not improve trial enrollment. Her takeaway was blunt: “It's important to publish negative trials.” (Fumiko Ladd Chino). Arturo Loaiza-Bonilla’s related thread pointed toward “collaborative intelligence” and Moravec’s Paradox rather than full automation (Arturo Loaiza-Bonilla).

For CME providers, this is a curriculum-design problem. Generic AI literacy can explain model types and risks, but it does not teach a clinician what to do when an ambient note is mostly right, a trial-matching alert is operationally weak, or a model’s answer looks plausible but may not transfer across settings. This extends an earlier brief on clinicians building their own AI tools while CME still teaches literacy: the demand is now for repeatable human-AI routines, not just comfort with the technology.

The implication: AI education should include short, specialty-specific rehearsals where learners write prompts, inspect outputs, identify missing context, and decide what remains a human responsibility.

Surgical teaching is becoming a workflow skill

At AUA25, the education signal was different but related: clinicians were treating teaching itself as something that needs structure, tools, and documentation.

One urologist highlighted the Briefing-Intraoperative teaching-Debriefing model and wrote, “Briefing, Intraop Teaching and Debriefing for all surgeries should be the norm during training!” (Chandru Sundaram). Another AUA25 thread praised an “Educating the Educator” course, pointed to the BID model for operative cases, and emphasized formative and summative feedback, letters of recommendation, and documenting teaching, education, and mentorship as academic accomplishments (Seth Cohen).

This is not just a urology note. Urology is the source context, but the provider implication is broader for procedural specialties: faculty development works better when it maps to the actual teaching sequence. Before the case, define the learner’s goal. During the case, use focused teaching scripts. After the case, debrief with reflection, reinforcement, correction, and documentation.

For CME teams, the opportunity is not another lecture on mentorship. It is a portable educator-skills package: BID scripts, feedback rubrics, sample debrief language, and academic-portfolio templates that departments can reuse. The question is whether your educator programs help faculty teach in the moment where teaching happens, or only describe good teaching after the fact.

What CME teams should audit this week

• Replace broad AI overviews with case-based exercises that pair prompting, output appraisal, and bias checks.
• Measure AI education by documented workflow changes or review habits, not only pre/post knowledge gains.
• Package procedural faculty development as before-during-after teaching tools, with feedback rubrics and CV documentation templates.

What to reconsider

If your catalog still separates “AI literacy” from clinical workflow, or “faculty development” from the operative teaching moment, this week’s signal argues for a tighter design standard. Clinicians are asking less for explanation and more for guided practice: show me the tool, the prompt, the feedback exchange, the debrief, and the judgment call I need to make when the real case is moving.