Test environments
Artificial intelligence evaluation

The suitability of intelligent systems lies above all in their ability to adapt (a conventional system, on the other hand, is judged quantitatively on its performance within a perfectly delimited and specified operating domain) . So it's not just a question of objectively quantifying function and performance, but also of validating and characterizing operating domains which are by nature variable, often strongly so, particularly in the case of so-called "open" environments....