IA on the job: what still needs to be learned when answers remain accessible?
AI can generate an explanation whenever it is requested, without requiring the user to reconstruct it. The managerial question is specific: how can essential knowledge still be learned when formulating it can be delegated?
What the literature establishes
Betsy Sparrow, Jenny Liu and Daniel Wegner (2011, Science) showed in experiments on remembering information that expecting future access can change what participants retain. Under some conditions, they remember less of the information itself but more about where to find it. These findings describe a reorganisation of memory in response to an external resource, not a general deterioration in intellectual abilities.
Finding an answer is not reconstructing it
Henry Roediger and Jeffrey Karpicke (2006, Psychological Science) showed that recalling studied material can support delayed retention more effectively than rereading it. Their experiments thus distinguish ease of access during study from the ability to recall the content later. For workplace AI, this suggests a hypothesis to test: regularly receiving an explanation does not necessarily replace the effort of reconstructing it.
The premature conclusion
One might conclude that AI should be restricted to make employees memorise more. These studies justify neither that general rule nor the idea that all professional information should be retained. The more relevant decision concerns the knowledge people need to have readily available themselves to act, understand an exception or recognise that an answer requires verification. (our executive and employee training programmes)
What the evidence cannot establish
These studies address neither today's generative assistants nor lasting learning within professional teams. They cannot establish an optimal frequency of AI use, and recalling an explanation does not prove that someone can apply it. Nevertheless, practices that measure only how quickly an answer is obtained overlook what the person can subsequently draw on without assistance.
A practical check in Lausanne
Within « IA on the job », SHR could invite a Lausanne team — in medtech, higher education, an international sports federation or a Lake Geneva region scale-up — to select a professional rule it considers essential to master without assistance. After working with AI on that rule, a delayed exercise, without tools and using a new fictional case, asks participants to explain the rule and then apply it; a predefined rubric distinguishes correct recall from correct application. The measure is the proportion of participants meeting each criterion, with a comparable exercise before the learning activity providing a baseline. This check does not demonstrate a causal effect of AI, but makes the gap between an obtained answer and usable knowledge observable, without exposing confidential data. To go further: explore the AI on the Job training in Lausanne, or browse our executive and employee training programmes in Switzerland.
In pictures: AI on the Job in Lausanne
- ai on the job
- Lausanne
- research
- training lausanne