IA on the job: allocating assignments without confusing balance with fairness

An AI assistant can suggest how to allocate cases, clients or projects based on reported availability. The managerial question is specific: how can managers check that this allocation does not concentrate learning opportunities among the same people under the guise of balancing workloads?

What the literature establishes

Jason A. Colquitt (2001, Journal of Applied Psychology) provides empirical support for four distinct dimensions of organisational justice: distributive, procedural, interpersonal and informational. The outcome, the process used to produce it, the treatment of people and the explanations provided therefore do not form a single undifferentiated judgement. Applied to AI-assisted allocation, this distinction cautions against treating equal volumes of work as sufficient evidence of fairness.

Dimensions of justice that cannot replace one another

Jason A. Colquitt, Donald E. Conlon, Michael J. Wesson, Christopher O. L. H. Porter and K. Yee Ng (2001, Journal of Applied Psychology) show, in a meta-analysis, that these dimensions have partly distinct relationships with workplace attitudes and behaviours. A detailed explanation and a favourable allocation are therefore not straightforward substitutes. For managers, this means examining allocation criteria separately from the value of the assignments allocated.

The premature conclusion

One might conclude that an allocation becomes fair as soon as its criteria are explicit and applied consistently. That would be premature: a criterion such as previous experience may help ensure an assignment is handled reliably while persistently reserving developmental assignments for people who are already experienced. This risk is a hypothesis to test within the organisation, not an effect of AI established by the cited studies. (our executive and employee training programmes)

What the evidence does not establish

These studies concern organisational justice, not the allocation of assignments by generative AI. They can neither certify a tool's fairness nor determine the appropriate weighting of availability, competence and professional development. Any assessment must also consider expressed preferences and actual constraints: a high-profile assignment is not automatically a welcome opportunity.

A practical check in Lugano

Within SHR's « IA on the job » programme, an exercise could examine all assignments eligible for assisted allocation within a team over one month: banking cases, fashion launches, supplier relationships in trading or projects in an Italian-speaking family business. Before allocation, the manager and team would define in Italian what constitutes a development opportunity, then record the eligible people, their availability, their preferences and the final allocation for each assignment. The measure would be, for each person, the number of developmental assignments received divided by the number for which they were eligible, available and willing. A gap would not prove unfairness, but would provide a verifiable basis for examining allocation decisions, without sending client data to an unauthorised tool. To go further: explore the AI on the Job training in Lugano, or browse our executive and employee training programmes in Switzerland.

In pictures: AI on the Job in Lugano

• ai on the job
• Lugano
• research
• training lugano