Being resigned to something is not the same as wanting that thing.
The outrageous prices of used (low telemetry) cars make them a poor value proposition, for example. Depending on the model, a used car can be a decade old with half of its lifetime miles on it already and the seller will still want $14k or something crazy like that. When you factor in the parts of its history that carfax doesn’t cover and the stream of repairs you’re likely to be saddled with, finding a 0%-2% financing deal on something new at ~$20-24k begins to look a lot more attractive.
There was a comment made here a month or two ago, about someone from a SE Asia country (I forget which; maybe Bangladesh?) talking about how their government installed AI traffic enforcement cameras all over, and they went from one of the deadliest, most chaotic road systems in the world to Singapore-level rule-following in the space of just a couple months.
As a result, the cameras have become quite popular because there's less traffic, travelling is easier and more predictable, and road deaths have plummeted.
I imagine there are a number of similar situations in every country, where the deployment of AI surveillance has a sufficiently measurably positive impact that it becomes popular.
What does the AI actually do? Identify specific cars and their drivers, or something? What actual enforcement actions were taken as a result, and against how many drivers before it stopped?
One might wonder how much of the effect was placebo (i.e., people conforming simply due to the belief of being surveilled).
Yes, exactly. It identifies cars and drivers who are committing traffic violations, and issues citations.
I don't have the specific numbers you ask for, seeing as I can't even remember the country, but it's not difficult to imagine that if sufficiently many such cameras are installed that most moving violations on major roadways are caught and punished, then yes, people will come to (reasonably) believe that the cost of committing the violation is close to the price of the fine, rather than close to zero.
It's one thing to blow by a pedestrian crossing the street, it's another to do so knowing you'll probably be charged $150 (or whatever) for having done it.
In this case, incentives are for people who'd like to reduce traffic deaths.
If you read "one can see how someone might not object to this" and understand "this is a good thing to have" then I'm afraid the issue is you contributing the stats that news agencies like to publish about American reading comprehension trends.
It's a trained model, just like Claude.
Your choice to see "AI" and assume some narrow definition is a personal issue that you could correct on your own if you chose.
We're in a thread about AI surveillence.
Note the "I imagine there are a number of similar situations in every country, where the deployment of AI surveillance has a sufficiently measurably positive impact that it becomes popular"
Someone said please don't refer to license plate scanners as AI. One can be obtuse and take that at face value, or one could read it as "please don't lump more deterministic ML together haphazardly by calling them all AI. "
This is what I mean by broad and narrow. You could have said technological surveillence and it had the same effect.
We're only in a thread about AI surveillance because I used the work "AI" when describing the license plate cameras.
Neither the comment I replied to, nor the original article, are particularly about AI in general nor about LLMs specifically. They're about data collection. The "AI" topic is a tangent.
I don't see the difference between "please don't lump more deterministic ML together haphazardly by calling them all AI" and taking what was said at face value. I completely understand the commenter's request that I encode their values into my speech in the way they prefer! I just disagree with them and decline to do so. What you misread as me being obtuse and not "reading" it correctly, is in fact just me saying "no".