So... between my gsgw fixation I've managed to finish reading The Reverse Centaur's Guide to Life After AI by Cory Doctorow, and I think it has a lot of really useful insights by viewing AI through an economic and labour lens.
The book isn't inherently anti-AI, but rather, anti-AI-bubble. And does a good breakdown of what the AI bubble is built from, who stands to benefit, and how to properly critique it.
Some salient quotes (which might be paraphrased cause I listened to the audio version and don't have it open).
The important thing about a technology is not what it does, but who it does it for and who it does it to.
An important focus of the book is how technology make workers reverse centaurs. A centaur is what you would call a human who employ technology to complete their task better. A person using assistive technology. In the case of AI, a writer who may ask the AI to run a search for a quote they vaguely recall. A reverse centaur, in this case, is a human who's an extension of a machine so the machine can complete its task. In this case, a writer who is given a workload impossible to complete by human and is therefore forced to review works output by AI and take the fall for any "human error", or an Amazon worker that has to package things at rate set by an algorithm.
Cory argues that based on whether workers are subjected to AI as centaur or reverse centaurs, their experience are vastly different.
This is "who it does it to".
As for "who it does it for", there's bosses and stake holders. Here's the key points:
Bosses hates paying workers.
AI companies convinces its stakeholders, who are mostly bosses, that AI can replace workers for cheaper, if only you dump in more money to develop it. This is a lie.
But it's a lie bosses are willing to believe in, because AIs cannot withhold its labour for moral reasons or demand better compensation like human workers.
Even if AIs cannot replace human labour, as long as bosses can be convinced of this lie, workers will be fired, and services will get worse.
Criticism of AI that focuses on the threat of it becoming a "super-intelligence" are actually hyping it up.
AI sector drums up this hype by tauting about how many jobs it can replace, even if these jobs, as important as they are to the workers, are mere pennies in terms of expense. Our paranoia about being replaced only drums up the hype.
Bosses wants it to replace jobs. And AI becoming a super-intelligent death-ray hurts no one's bottom line (which is why AI bosses loves it when the focus is drawn to its hypothetical "existential-threat", it is a distraction from the real harm its causing)
AI being a pointless money-sink does.
The book also talks about consequences of the AI bubble popping and makes an argument for open-source localized AIs (the type that can be ran on a laptop), which doesn't need 50 data centers to run.
I think if you're already staunchly anti-AI, there'll be some points that you may disagree with. Nevertheless I think it's a very insightful work to read with an open mind.