AI assistants: does a polished interface demonstrate quality in Basel?
No-code tools make it possible to give an AI assistant a professional appearance very quickly. For a team manager, the question is specific: how can we avoid choosing the most appealing prototype rather than the most useful one?
What the literature establishes
Richard Nisbett and Timothy Wilson (1977, Journal of Personality and Social Psychology) show that an overall impression can alter assessments of particular attributes without people correctly recognising that influence. Their experiment concerns the evaluation of an instructor, not a digital tool. Nevertheless, it establishes a problem relevant to assistant demonstrations: judgements about several qualities do not necessarily constitute independent assessments.
Ease of reading can become a quality signal
Rolf Reber, Norbert Schwarz and Piotr Winkielman (2004, Personality and Social Psychology Review) offer a theoretical review linking processing fluency to aesthetic pleasure. A presentation that is easier to process can be appreciated for that very ease. Applied to assistants, this mechanism suggests a hypothesis to test: the readability of a response might influence an assessment intended to measure its operational usefulness.
The premature conclusion
It would be excessive to conclude that an attractive interface is suspect or that a plain prototype is more reliable. Readability is itself a quality when it helps people locate information or understand a proposed action. The useful distinction is therefore between measured ease of use and an overall impression treated as evidence. (our executive and employee training programmes)
What a demonstration cannot tell us
These studies measure neither the performance of current AI assistants nor the size of any presentation effect in a Basel company. Nor can a successful demonstration isolate the contributions of content, interface and the chosen scenario. An overall satisfaction score leaves these dimensions entangled and is insufficient for a deployment decision.
A practical check in Basel
As part of SHR's programme « Créer ses assistants et agents IA sans coder », a Basel team working across pharmaceuticals, life sciences, Rhine logistics and international R&D can test an assistant that prepares sample-shipping requests using fictional cases. Present identical responses in two formats, plain and elaborate, assigning versions randomly and ensuring that nobody sees the same case twice. Measure reported appeal separately from the proportion of required fields correctly transferred to a request form, using a predefined scoring guide. If presentation improves appeal alone, the result documents an interface preference, not an improvement in the work completed. 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