Accountability Theater - Again
The actual monitoring (data capture) is not (or need not be) the issue. The issue is human access to - for whatever purpose - the captured data. Surveillance is merely one of the purposes.
The problem, as ever, is that we the people have no means of determining when the technology and relevant data is switched from one mode (capture) to the other (access). This is the gap I've previously and repeatedly referred to as "Accountability Theatre".
The short version is that there are always going be justifiable cases for surveillance. For a traffic based example, I rather like the idea that we could develop an AI system to watch motorway traffic in order to identify genuinely dangerous situations emerging in real time. This would include things like some imbecile driving the wrong way, or a car randomly weaving in a manner likely to indicate someone falling asleep at the wheel. Or someone driving at twice the speed limit in busy traffic. Where such threats are identified, the AI prods a Human and they can raise an alarm, send out a traffic cop, turn on the warning signals etc etc.
I doubt that anyone is going to argue with that kind of use of surveillance. Where it crosses the line into authoritarianism is, for example, with John Robson's suggestion that it could also be used to "enforce appropriate speed limits". This is a grey area. Certainly, as hinted above, some speeds would qualify for the alarm surveillance mentioned above . Driving at 80 on a reasonably clear motorway does not. That said, we should have no issue with the data captured being used, after a serious incident (eg a fatal accident) to see to what extent either speeding or careless driving contributed to the accident.
And in all cases, where data is accessed for any reason whatsoever, by a human actor, it should not be technically possible for such access to take place, without it being subject to the most rigorous surveillance of all, with the data being provably captured to an immutable audit trail.