Two people using the same model can get very different results. To understand why, we need to look at prompting technique alongside domain knowledge, problem framing, critical thinking, written communication, data literacy, and judgment about quality.
An expert knows the criteria for a good output. They notice missing context, catch incorrect assumptions, and reject an answer when necessary. A creative thinker can broaden the range of options. A strong communicator clearly describes the purpose, constraints, and target audience. AI can magnify the effect of these capabilities, but it can also magnify the way gaps show up in the output. I think it is important to account for both when designing training.
Corporate training can make room for work on real cases alongside ready-made prompt lists. Having employees define the problem, form an independent view, critique an AI output, and justify their decision reveals what they know and how they think. AI literacy can be built on this foundation.
To establish a baseline, human capabilities and quality criteria can be defined by role. Assessment through real cases can combine self-assessment, peer perspectives, and expert feedback. Training can then be designed around the capability gaps observed.
Frequency of use alone is not enough to measure progress. How an employee thinks, the quality of their output, and the reasoning behind their decision also need attention. The effect of training on work can be assessed through better decisions and business outcomes.
Ask yourself
- Do we have the domain knowledge to recognize a good output?
- Which of our strengths does AI amplify, and which gaps does it reveal?
- Does training increase usage or the quality of work?
