An old parable goes like this: A client hires a plumber to fix a leak. After a few hours, the plumber taps one rivet with his hammer, and the leak is fixed. The plumber charges $10,000. The client asks, "You expect me to pay you $10,000 for that?" The plumber says, "I charged $1 to physically hammer it, and $9,999 to know where to hammer."
Think about the last time you hired someone to fix your car, or a leak in your house. Were you paying them to physically turn nuts and bolts? Or were you paying them for their expertise and problem-solving abilities? Every time there's a post about AI and jobs, there are tons of people suggesting that knowledge work is becoming obsolete, and therefore people should go to a trade school for a blue-collar job. Are people somehow forgetting that in a trade school, you learn things about how to do your job more effectively, which is, in effect, knowledge work?
If the day comes that AI can completely automate a typical tech worker's job, it won't be very long before it can automate the job of a plumber, electrician, mechanic, or repairman. Especially with multi-modal models, soon all you will have to do is point your camera at something and ask it where you should point your camera next, which part of the house you should show it to have it get a better diagnosis, what kind of test to run to figure out what's wrong with your car, or what tool you should buy or how you should physically turn the wrench.
Note: We are assuming AI in this hypothetical future had already become powerful enough to replace a typical software engineer or tech worker. As long as AI is not capable of the task I described above, then it is also not yet capable of replacing typical white-collar "knowledge work" in the first place.
