Whistleblowers: Software keeping inmates in Arizona prisons beyond release dates

That's hard to do because all of those systems are intertwined. If you use the Montana DMV program, and you want people who get DUIs to have their driver's license suspended, now you have to use the Montana Penal System program. Except Montana's Penal System has a bunch of exceptions written into it for laws that can be either a misdemeanor or a felony, and they don't allow any time off for good behavior. So now are you going to adopt Montana's laws too? There's tax code stuff in there too, so I guess we're lumping in the Montana tax system as well.

I think the problem is that unlike our more notable branches, we don't hire experts in the field. I don't mean they're incompetent at technology, but that a problem like this really exists at the intersection of government and technology. We keep hiring general-purpose contractors to build things like this, and then we're shocked when it falls apart in the environment governments exist in.

We need companies that specialize in this intersection. Companies that can keep public sentiment in mind and build an architecture that's flexible in the places where society is. It's the same way that most of us in general purpose IT try to build systems that can adapt to changes in the IT landscape. Put it in Docker so we can run it on a cloud, on bare metal, on k8s and probably on whatever's next. Governments struggle to pivot like that due to funding (how do you argue for funding for features since you can't earn revenue?), and because a lot of it is legislated out of their control. Learning to read the public sentiment is just like us reading trends in a newsletter.

> humans have to deal with the consequences of "dumb" actions (e.g. by looking through their spam folders for false positives)

Email programs generally have a mechanism for reviewing email and changing the classification. I think your "pure-AI" phrase describes a system that doesn't have any mechanism for reviewing and adjusting the machine's classification. The fact that a spam message winds up in your inbox sometimes is probably that low-confidence human-in-the-loop process we've been talking about. I'm sure that the system errs on the side of classifying spam as ham, because the reverse is much worse. Why have two different interfaces for reading emails, one for reading known-ham and one for reviewing suspected-spam, when you can combine the two seamlessly?

Perhaps you've confused bad user interface decisions for bad machine learning system decisions. I'd like to see some kind of likelihood-spam indicator (which the ML system undoubtedly reports) rather than a binary spam-or-not, but the interface designer chose to arbitrarily threshold. I think in this case you should blame the user interface designer for thinking that people are stupid and can't handle non-binary classifications. We're all hip to "they" these days.