The author has two points that continually distract from each other. They have a specific critique of worker conditions within tech companies, but also a broad

The author has two points that continually distract from each other. They have a specific critique of worker conditions within tech companies, but also a broad skepticism that AI can really be all that effective or innovative, rather than an abstraction over the workers who "really" do the work. Their first point is solid and is accompanied by immediate calls to action, their second is vague, wrong, and potentially dangerously naive. They use AI, "so-called AI", or "AI" (with scare quotes) interchangeably. Besides being unnecessarily confusing with their own terminology, they are bringing in a pointless philosophical debate about what "intelligence" or "learning" really is. The "40% of AI startups that don't actually use any AI" is mentioned not because they're misleading investors or users, but because they point out that the "AI" label makes workers less visible, and then they can tie that to Amazon's "artificial artificial intelligence" and go on from there to show or at least imply that pretty much all AI is a big smokescreen to justify the marginalization of workers. However... isn't that begging the question? Do we actually care whether these systems learn and think the same way humans do, do their own original work, or whether they're just a worker pretending to be a chatbot? Does that make it okay to underpay gig workers, give them PTSD, or subject them to poor conditions? Maybe "real AI" never materializes, but also maybe it just keeps improving until it doesn't matter. If so it's dangerous to deny that these systems can ever meaningfully replace or augment humans. We can't pretend that we live in a universe where something will forever be inferior to good old-fashioned human labor just because of the disturbing social implications.