Training with AI: does completing the exercise mean learning?
AI can support employees during training by explaining a procedure or suggesting an answer to a practical case. The managerial question is specific: how can assisted success on an exercise be distinguished from the acquisition of reusable knowledge?
What the literature establishes
Henry L. Roediger III and Jeffrey D. Karpicke (2006, Psychological Science) compared repeated study of texts with conditions involving recall tests. Repeated study supported performance in the very short term, whereas retrieval practice produced better retention on delayed assessments. Ease during learning and the subsequent availability of knowledge are therefore not interchangeable indicators.
Retrieving knowledge is not simply recognising it
Jeffrey D. Karpicke and Janell R. Blunt (2011, Science) showed that retrieval practice could produce more learning than elaborative studying with concept maps, including on questions assessing comprehension and inference. In this setting, reconstructing knowledge from memory offered a benefit that organising visible content did not provide to the same extent. For AI-supported training, this suggests reserving moments when participants must formulate their own reasoning before consulting an answer.
The premature conclusion
One might conclude that AI should be banned during training. Yet these studies do not compare training with and without AI and do not support that conclusion. The issue is instead to distinguish moments of explanation, when assistance may be useful, from moments of retrieval, when the aim is to observe what a person can draw on independently. (our executive and employee training programmes)
What a successful exercise cannot establish
These experiments address bounded learning materials, not the full range of managerial competencies or medtech procedures in real working conditions. They establish neither the optimal frequency of unaided exercises nor their effect on professional performance with an AI tool. An unaided assessment provides information about available knowledge; it does not replace an assessment of work under its usual conditions.
A practical check in Lausanne
As part of SHR’s « Manager à l'ère de l'intelligence artificielle » programme, a Lausanne-based team from medtech, higher education, international sport or a Lake Geneva region scale-up can test a bounded learning objective using a fictional, non-sensitive case. Following an AI-supported exercise, introduce a new case of comparable difficulty the next week, first without assistance and then with the tool, using the same scoring rubric defined in advance. Measure separately the proportion of criteria met without help and then with assistance, and document the steps that still require support. This check does not establish a causal effect of AI, but it avoids equating the quality of an assisted answer with knowledge that is actually available. To go further: explore the Managing in the Age of Artificial Intelligence training in Lausanne in the canton of Vaud, or browse our executive and employee training programmes in Switzerland.
In pictures: Managing in the Age of Artificial Intelligence in Lausanne
- managing in the age of artificial intelligence
- Lausanne
- research
- training lausanne