Communication Gap Observation

Systematic breakdowns in communication between AI systems and users, and between AI teams and stakeholders. AI outputs are often misinterpreted because of unstated assumptions.

The Gaps

AI → User:

• Model uncertainty not conveyed
• Assumptions not surfaced
• Limitations not visible
• Confidence conflated with correctness

AI Team → Stakeholders:

• Technical capabilities oversimplified
• Limitations not communicated
• Benchmark performance mistaken for production readiness
• Failure modes not anticipated

Consequences

• Users trust AI inappropriately (over- or under-reliance)
• Stakeholders have misaligned expectations
• Failures surprise everyone
• Trust erodes when reality differs from communication

Bridging Strategies

• Explicit uncertainty communication
• Worked examples of limitations
• Expectation-setting documentation
• Regular calibration conversations

Related: 05-atom—uniform-confidence-problem, 01-molecule—appropriate-reliance-framework