simulated user

Terms from Artificial Intelligence for Human Computer Interaction

A simulated user uses some form of AI, for example an LLM with a crafted prompt, or other computational model to emulate the behaviour of a user. This might be used as part of evaluation or even to automatically fill in a questionaire during 'user' research. There are obvious problems in by-passing real user engagement but, just as with using expert evaluation, also times when it may be advantageous in terms of rapidity of feedback. In some cases, there may also be accessibility advantages compared with real user evaluation, as simulations can deal with multiple kinds of intersecting abilities.
While substantial use of simulated users is recent, there is a history of research dating to the earliest days of HCI, notably cognitive complexity theory, which used a cognitive model of the user. Another area that has been used heavily in practice are various tools to emulate different foms of visual differences incuding forms of colour blindness and macular degeneration.

Used in Chap. 6: pages 84, 89

Also known as: surrogate user, synthetic user