Since changes made to the model by the AI may not be implemented as intended, and the results may be difficult to understand, it would be helpful if the AI’s capabilities could be limited—
for example, to workflow guidance and information derived from the model
However,
modifications to the model—such as changing parameters, creating elements, or loading families—should be excluded.
It should generally be possible to control this, either in the Assistant’s settings, the options menu, or through the administrator (as with Add-Ins).
Translated with DeepL.com (free version)