To be fair, wouldn't that suggest strong oversight might work then? True, any test might differ from real-world conditions, but theories need to be tested one way or another and it would provide some evidence.
While caution during early testing might lead to less misuse, one could also imagine countervailing factors. For example, lack of familiarity with a new technology might lead to might lead to mistakes. Regulations are written in blood, as they say, and the development of new ethical guidelines may take time.
Which, as we've noted, could be a pragmatic reason to let others be the test subjects. I'm not eager to open the can of worms myself, though it might feel a bit selfish to put it that way.
Fair point, that might work if: 1. a public safety/citizens oversight committee does its job consistently, 2. isn't loaded with police-friendly stooges 3. and isn't gradually de-fanged over time in terms of its power.
All three things, with time, can be manipulated by any given city hall, which is often lock-step with the police force.
"...but theories need to be tested one way or another and it would provide some evidence"
Agreed. And I say let's look at how they've deployed facial recognition in China to put those theories to bed.