In Lugano, do we judge an AI assistant differently when we built it ourselves?
The programme “Creating your own AI assistants and agents without coding” opens up an activity previously reserved for specialists. This raises a managerial question: how should an assistant be evaluated when its designers are also the people recommending its use?
What the literature establishes
Michael Norton, Daniel Mochon and Dan Ariely (2012, Journal of Consumer Psychology) show, in experiments involving objects assembled by participants, that the work of making something can increase the value assigned to the result. This “IKEA effect” depends in particular on successful task completion: effort alone does not consistently produce this increase in valuation. These studies establish an effect on the evaluation of objects, not on the quality of AI assistants.
When the tool becomes “our” tool
Jon Pierce, Tatiana Kostova and Kurt Dirks (2001, Academy of Management Review) propose a theoretical framework of psychological ownership: individuals may feel that an object belongs to them without legally owning it. In this framework, exercising control, developing intimate knowledge of the object and investing oneself in it provide routes towards that feeling. Configuring an assistant’s workflow and testing its responses could bring these conditions together; this is a possible application to investigate, not an established finding about AI.
The premature conclusion
One might conclude that designers should be excluded from evaluation. That would confuse a risk of subjective valuation with a lack of competence: the people building the tool also understand the constraints it must meet. A reasonable precaution is to distinguish their contribution to defining the criteria from the final judgement of the outputs. (our executive and employee training programmes)
What these studies cannot settle
Experiments involving assembled objects cannot establish the size of any corresponding effect in a bank or an SME, and the psychological ownership framework is not a causal trial involving assistants. A designer’s preference may also reflect a better understanding of the use cases. An enthusiastic internal demonstration therefore cannot distinguish attachment to the tool from its actual suitability or the advantage of knowing how it works.
A practical check in Lugano
As part of the programme, a team in Lugano can select an Italian-language task using fictional cases suited to its business: an administrative reply in banking, a product description in fashion, a supplier enquiry in trading or an order follow-up in an Italian-speaking family SME. It compares its assistant’s outputs with those produced by the usual method, using the same cases, criteria set before testing and provenance concealed as far as possible. Designers and colleagues not involved in development separately assess the proportion of outputs usable without substantial revision; the gap between their preferences for the assistant is the result to examine. That gap does not prove an attachment effect, but it makes it possible to investigate a question that designer satisfaction alone leaves unanswered. To go further: explore the Building AI Assistants and Agents Without Coding training in Lugano in the canton of Ticino, or browse our executive and employee training programmes in Switzerland.
In pictures: Building AI Assistants and Agents Without Coding in Lugano
- building ai assistants and agents without coding
- Lugano
- research
- training lugano