Topic 01
AI literacy is more than prompt training.
Prompt training teaches people how to ask. It does not teach them how to judge. This one is about the part that stays human: knowing which sources count, noticing when a confident answer is quietly filling a gap, and deciding how much checking the work has earned.
- Why fluency is not evidence, and how plausible errors get past busy people
- Grounding: giving authoritative sources control over inference
- Verification matched to consequence rather than applied uniformly
- The four Operator disciplines: Frame, Ground, Verify, Govern
For district and agency staff, instructional leaders, and professional learning teams
Topic 02
AI in the public square.
A weak internal draft can be corrected. An unsupported answer sent to a family, or quietly wired into a workflow, becomes an institutional action. This one is about what has to be true before an agency lets that happen.
- Hesitation is not resistance: why the hardest questions are worth listening to
- Expert in the loop, not a generic human in the loop
- Governing the consequence rather than the novelty of the tool
- What must be true before an agency relies on an answer
For agency leadership, boards, policy staff, and conference general sessions
Topic 03
The data headache.
Most agencies already have data. That is not the same as having knowledge. The headache starts when definitions disagree, ownership is unclear, and nobody can say which report to trust. It does not go away by buying another tool.
- Why the hardest failures live between systems and people, not inside either
- People, process, policy, platform, and the purpose test that governs all four
- Provenance before intelligence: why AI cannot rescue ungoverned data
- Enterprise capability sized for real agency budgets and staffing
For data teams, CIOs and CDOs, state and regional agencies, and technical tracks
