Designing AI for people who hate it.
Frontline workers, clinicians, analysts — the people who actually use enterprise AI often start out skeptical. Three principles that make the skeptics into advocates.
- Earn trust per interaction. Every output is a chance to keep or lose them.
- Never make work. The moment AI adds steps to their day, you've lost.
- Fail visibly. Hidden failures destroy trust slower but more permanently than visible ones.
Earn trust per interaction.
A skeptical user gives you about three interactions before they decide whether the system is worth their time. Every output is a small bet. Confident wrong answers are devastating. Hedged correct answers are forgettable. Confident correct answers, sourced and explained, build the trust that lets the system survive its inevitable mistakes.
Design the output as an interaction, not as a deliverable. Let the user verify, correct, and inspect. Show your work.
Never make work.
The moment AI adds a step to someone's day — even one — the cost-benefit calculation flips. AI that adds a click is dead on arrival. AI that removes a click is grudgingly tolerated. AI that removes a workflow is loved.
Test this rigorously. Watch a real user do their work with and without the AI. If the AI makes the path longer, more steps, or more cognitive load, redesign it.
Fail visibly.
The worst AI failures are the silent ones: confidently wrong outputs that the user trusts and acts on. Visible failure — "I don't know," "I'm not confident here," "this is outside my training" — is uncomfortable for the AI to produce but builds long-term trust.
Your skeptical user already expects the system to fail. They're impressed when it admits it.
What skepticism actually sounds like.
If you're designing for people who hate AI, you have to know what they say when they hate it. Real lines we've heard from real users on real engagements:
“I'm not pasting patient notes into someone's chatbot. End of conversation.”
“If it's wrong about the policy and I act on it, that's my license, not theirs.”
“The last 'AI tool' they rolled out added 20 minutes per shift and we all turned it off.”
“Show me where it got that. I'm not signing off on a sentence I can't trace.”
Each of those lines is a design requirement. Tenant isolation. Citations on every answer. Time-to-value measured in seconds. Auditable lineage on every recommendation. Build for skeptics first; the enthusiasts will use anything.
BizzSoftware designs, builds, secures, and runs the internal applications your teams work in every day — with AI features built in. About us →