AI assistants: should their use be rewarded?
After building a no-code assistant, a manager may be tempted to link its use to a bonus or an individual target. The management question is specific: should frequency of use be rewarded, or should the appropriateness of using the tool be assessed?
What the literature establishes
Edward Deci, Richard Koestner and Richard Ryan (1999, Psychological Bulletin) synthesised experiments on the effects of external rewards on intrinsic motivation. Their meta-analysis indicates, in particular, that expected tangible rewards contingent on engaging in or completing an activity can reduce subsequent freely chosen engagement in that activity. It distinguishes these situations from positive feedback, which can support interest: different forms of recognition do not necessarily have the same effect.
When incentives change the meaning of a rule
Uri Gneezy and Aldo Rustichini (2000, The Journal of Legal Studies) studied the introduction of a fine for parents arriving late at childcare centres. Late arrivals increased after the fine was introduced, and removing it did not restore the previous level during the observation period. The authors suggest an interpretation: a financial penalty can turn a social obligation into what is perceived as a paid service.
The premature conclusion
These findings do not demonstrate that a bonus linked to AI assistant use would necessarily be counterproductive. Instead, they invite scrutiny of what the incentive makes desirable: solving a work problem or displaying usage that meets a target. Recognition based on an appropriate use case is not equivalent to a reward based on the number of conversations opened. (our executive and employee training programmes)
What these data cannot establish
Neither study examines no-code AI assistants, and a fine for lateness is not a technology adoption bonus. They therefore cannot predict either the magnitude or even the direction of an effect in a Neuchâtel company. Nevertheless, the common practice of tracking logins remains insufficient: it measures technical activity, not the reason that activity was chosen.
A practical check in Neuchâtel
As part of SHR’s programme « Créer ses assistants et agents IA sans coder », a Neuchâtel team working in microtechnology, watchmaking or precision microelectronics could test an assistant that prepares team handovers, without linking the pilot to individual bonuses. Over a period defined in advance, each eligible handover would be recorded along with whether the assistant was used, the reason for that choice, and whether the receiving team accepted it or requested revisions. The main measure would be the proportion of handovers accepted without revision, reported separately with and without the assistant while accounting for the type of case. This monitoring would not prove a causal effect, but it would help check whether observed usage accompanies a useful outcome rather than merely an increase in logins. To go further: explore the Building AI Assistants and Agents Without Coding training in Neuchâtel in the canton of Neuchâtel, or browse our executive and employee training programmes in Switzerland.
In pictures: Building AI Assistants and Agents Without Coding in Neuchâtel
- building ai assistants and agents without coding
- Neuchâtel
- research
- training neuchâtel