On rereading a proposed legal definition of AI, which I mentioned here, I noticed a few components that seemed difficult:
The defintion includes:
“In general, the more human-like the system within the context of its tasks, the more it can be said to use artificial intelligence” (page 3, line 22-25)
“Systems that think like humans” (page 4, line 1)
“Systems that act like humans” (page 4, line 3)
“that exhibits its apparently intelligent behavior at least as advanced as a person” (page 4, line 18)
A problem here is that if a judge were to want to apply this definition to determine if something is artificial intelligence, he or she would have to make a lot of determinations about the standards at which humans act or think. References to the Turing test (page 4, line 4) are probably a lot more effective to create a usable definition than relying on analogies.