This would not have been a relevant question except for the rapid uptake and integration of AI language model capabilities in some very popular software platforms and applications. However, users don't seem to be offered a choice of whether to use the tool without these features, especially if they are built into applications that work with text and image.
There are some design patterns for using AI features that allow the user to know that these features are being applied, and to expect potential issues like hallucination or to treat the results differently. But I expect that for tasks that the AI model is sufficiently mature at dealing with, this will not be deemed necessary soon.
Are there examples of software applications that allow users to turn these features off (for whatever reason they may have), and how are they being implemented?
