It’s not a double standard. AVs aren’t human. We can set whatever standard we want for them. There’s no logical reason it should be consistent with the standard for humans.
The logical reason is that if they reach the point of being safer than human drivers but we ban them for not being perfect then lives are being lost unnecessarily.
First, the companies involved in self driving need to stop lying about their records. Tesla has engaged in years of making extremely suspect claims about their safety that do not stand up to scrutiny at all.
Self driving technology is much more scalable than a bad or negligent driver. So it is natural to be more concerned about it, even if you’re convinced that on average it has performed better than humans.