The claim that AI is a scam is a ridiculous and can only be stated by someone who doesn't understand the technology. Are we genuinely supposed to believe that capitalists hate profits and capital accumulation and are just wasting their money on something worthless? It's absurd. AI is already making huge breakthroughs in many fields, such as medicine with protein folding. I would recommend watching this video on that subject in particular. China has also been rapidly improving the speed of construction projects by coordinate them with AI.
To put it in laymen's terms, traditional computation is like Vulcans: extremely logical and have to go compute everything logically step-by-step. This is very good if you want precise calculations, but very bad for many other kinds of tasks. Here's an example: you're hungry, you decide to go eat a pizza, you walk to the fridge and open it, take out the slice, put it in the microwave to heat it up, then eat it. Now, imagine if I gave you just the sensory data, such as, information about what a person is seeing and feeling (hunger), and then asked you to write a full-proof sequence of logical statements that, when evaluated alongside the sensory data, would give you the exact muscle contractions needed to cause the person to carry out this task.
You'll never achieve it. Indeed, even very simple tasks humans do every day, like translating spoken words into written words, is something that nobody has ever achieved a set of logical if/else statements to replicate. Even something seemingly simple like this is far too complicated with far too many variables for someone to ever program, because everyone's voice is a bit different, every audio recording is going to have slightly different background noise, etc, and to account for all of it with a giant logical proof would be practically impossible.
The preciseness of traditional computation is also its drawback: you simply cannot write a program to do very basic human tasks we do every day. You need a different form of computation that is more similar to how human brains process information, something that processes information in a massively parallel fashion through tweaking billions of parameters (strengths in neural connections) to produce approximate and not exact outputs that can effectively train itself ("learn") without a human having to adjust those billions of parameters manually.
If you have ever used any device with speech recognition, such as writing a text message with spoken voice, you have used AI, since this is one of the earliest examples of AI algorithms actually being used in consumer devices. USPS heavily integrates AI to do optical-character recognition, to automatically read the addresses written on letters to get them to the right place, Tom Scott has a great video on this here on the marvel of engineering that is the United States Postal Service and how it is capable of processing the majority of mail entirely automatically thanks to AI. There have also been breakthroughs in nuclear fusion by stabilizing the plasma with AI because it is too chaotic and therefore too complex to manually write an algorithm to stabilize it. Many companies use it in the assembly line for object detection which is used to automatically sort things, and many security systems use it to detect things like people or cars to know when to record footage efficiently to save space.
Being anti-AI is just being a Luddite, it is oppose technological development. Of course, not all AI is particularly useful, some companies shove it into their products for marketing purposes and it doesn't help much and may even make the experience worse. But to oppose the technology in general makes zero sense. It's just a form of computation.
If we were to oppose AI then Ludwig von Mises wins and socialism is impossible. Mises believed that socialism is impossible because no human could compute the vastness of the economy by hand. Of course, we later invented computers and this accelerated the scale in which we can plan the economy, but traditional computation models still require you to manually write out the algorithm in a sequence of logical if/else statements, which has started to become too cumbersome as well. AI allows us to break free of this limitation with what are effectively self-writing programs as you just feed them massive amounts of data and they form the answer on their own, without the programmer even knowing how it solves the problem, it acts as kind of a black-box that produces the right output from a given input without having to know how it internally works, and in fact with the billions of parameter models, they are too complicated to even understand how they work internally.
(Note: I am using the term "AI" interchangeably with technology based on artificial neural networks.)