David Roth makes a very good point about how AI (whatever that means) will be implemented–not how it could be implemented, but how it will likely be implemented (boldface mine):

At this stage in this increasingly ubiquitous and increasingly janky technology’s strange simultaneous ascent and descent, AI is far more successful as a brand or symbol than it is as any actual useful thing. A technology this broken would not be something that government, for instance, would obviously want to use, given the extent to which a government’s credibility has traditionally been understood to depend upon being trustworthy and consistent, and the extent to which AI is currently unable to be either of those things. But if you didn’t really care about that, then you wouldn’t really care about that…

There is enough money and ambition behind all this stuff, and a sufficiently uncommitted collection of institutions standing between it and everyone else, that AI could become even more ubiquitous even if never actually works. It does not have to be good enough to replace human labor to replace human labor; the people making those decisions just have to go on deciding that it doesn’t matter.

It’s related to a point David Karpf made about digital butlers (boldface mine):

Agents, if they are developed at all, are going to be a bespoke, luxury good. They’ll be for discerning customers with money to spend on personalization. The business class lounge set. The rest of us will get info-sludge and degraded public services. That’s the status quo ante, at least. It’s what will happen if we don’t resist, and collectively demand a better future…

If you want to know where social technologies are headed, don’t focus on what the technology might be used for under ideal conditions.

Focus on the direction that currently-existing market forces will channel it.

Meanwhile, initial forays in using GhatGPT-4 for infectious disease consults have not gone so well (boldface mine):

Forty-four cases with a first episode of positive blood culture were included. ChatGPT-4 provided detailed and well-written responses in all cases. AI’s diagnoses were identical to those of the consultant in 26 (59%) cases. Suggested diagnostic workups were satisfactory (ie, no missing important diagnostic tests) in 35 (80%) cases; empirical antimicrobial therapies were adequate in 28 (64%) cases and harmful in 1 (2%). Source control plans were inadequate in 4 (9%) cases. Definitive antibiotic therapies were optimal in 16 (36%) patients and harmful in 2 (5%). Overall, management plans were considered optimal in only 1 patient, as satisfactory in 17 (39%), and as harmful in 7 (16%)…

The use of ChatGPT-4 without consultant input remains hazardous when seeking expert medical advice in 2023, especially for severe IDs.

(when you look at their scoring criteria, they were relatively lenient to boot).

To sum up, the various things referred to as AI will be adopted, based not on utility, outcomes, or social consequences, but whether or not those in positions of authority and power will want them adopted, regardless of the consequences.

Same as it ever was, I guess–but that’s not revolutionary in any sense.