AI assistants: should individual performance indicators be visible in Basel?
An assistant built without coding can generate indicators of turnaround times, corrections or requests handled by each employee. The management question is specific: does making these individual indicators visible to a manager improve the work, or mainly change the conditions under which it is performed?
What the literature establishes
John R. Aiello and Kathryn J. Kolb (1995, Journal of Applied Psychology) experimentally studied electronic performance monitoring and its social context. Their findings show that its productivity effects are not uniform, notably across skill levels, and identify a cost in stress. Making an activity observable is therefore not a neutral intervention.
Being observed changes the task
Robert B. Zajonc (1965, Science) proposes a classic mechanism of social facilitation: the presence of others strengthens dominant responses, which can help with a well-learned task and hinder a new or difficult one. The article concerns neither AI assistants nor digital dashboards. Nevertheless, it provides a reason not to expect visibility to have the same effect on a routine operation as on an uncertain scientific investigation.
The hasty conclusion
One might conclude that all individual visibility should be removed; these studies do not justify such a general rule. A record that allows an action to be reconstructed serves a different purpose from a ranking of employees by speed. In the programme “Créer ses assistants et agents IA sans coder”, this distinction becomes a design question: which records should be retained to understand the process, and which indicators should be displayed to evaluate people? (our executive and employee training programmes)
What the evidence cannot promise
These studies predate generative assistants and cannot quantify their effects in a Basel company. An electronic monitoring experiment and a social psychology mechanism do not demonstrate that a particular dashboard will impair R&D work. Conversely, an increase in closed requests does not establish an improvement if case difficulty, rework and outcome quality remain unknown.
A practical check in Bâle
In a Basel team coordinating pharma, life sciences, Rhine logistics and international R&D, test a no-code assistant on a non-critical workflow, such as preparing coordination requests. With participants’ agreement, compare two arrangements retaining the same necessary records: aggregate process monitoring and a named individual display for the manager, without using the pilot for HR appraisal. Randomly assign teams to the arrangements if numbers permit, then record the proportion of requests reopened because of missing information, median handling time and an anonymous stress rating. Establish before the test that a shorter handling time does not count as success if rework or stress increases; with few teams, treat the result as a local signal rather than causal proof. To go further: explore the Building AI Assistants and Agents Without Coding training in Basel in the canton of Basel-Stadt, or browse our executive and employee training programmes in Switzerland.
In pictures: Building AI Assistants and Agents Without Coding in Basel
- building ai assistants and agents without coding
- Basel
- research
- training basel