The Scenaptic.AI competency model for AI skills

What does "good with AI" actually look like? Most AI training never answers the question. You attend the sessions, you see the tools, you receive the certificate. What almost none of it tells you — or your employer — is whether you can now do anything differently. Ask what a programme's graduates are able to do, observably, that they could not do before, and the answers drift into the realm of "understanding", "discussing", "being aware of" — verbs that describe exposure, not capability.

This is not a new problem, and I did not meet it first in AI. I worked as a consultant in competency-based systems for pharmaceutical and health companies, and later brought that discipline to aviation — an industry that cannot afford to confuse exposure with capability. In work published by ICAO in 2010, I argued that a competency is not a topic covered but, in Spencer and Spencer's definition, an underlying characteristic of an individual causally related to superior performance in a job or situation — defined through observable behaviours and validated with data. Aviation defines skills that way because lives depend on the definition. The technology has changed since then. The question has not: how do you know someone can actually do the work?

Applied to AI, the question exposes a gap. Tool tutorials do what they promise: general functionality, taught the same way to everyone — useful, and by the same token incapable of creating an advantage, because an advantage cannot come from what everyone learns identically. Strategy education addresses the organisation. What neither provides, at any level, is the connection between AI and what you specifically do well: judging when to delegate a task to AI and when not to, evaluating an output against your own expertise, knowing how much to trust a result and what it would cost to be wrong. These capabilities apply to the analyst and the chief executive alike — an executive's judgment is expertise too — and almost nowhere are they stated as assessable outcomes. They are judgment skills. Nobody develops judgment by watching demonstrations.

The competency model we use at Scenaptic.AI defines AI skills the aviation way: observable behaviours, demonstrated in realistic simulations, assessed against evidence — because judgment is built by deciding, not watching. It is the same discipline behind the original article, applied to a technology that makes the difference between exposure and capability more expensive than it has ever been.

The 2010 ICAO Journal article, "Training the Digital Generation for the Aviation Industry," is available here.