But it is a niche and some are hedge fund guys or some crazy finance wizards, some do research or teach, while others are working at Starbucks because they couldn’t do something cool with it.
People here massively underestimate the technological curve we're on and how quickly tech jobs are, and will continue to be replaced in the coming years.
I think I commented a couple of years back how it won't be long before you can give AI a github issue and it will raise a PR for it automatically within minutes. Those types of tools are now available and pretty damn good. With the latest models they're now easily as good or better than developers with < 5 years experience. And for smaller more simple projects you probably don't need to hire devs anymore.
For now the latest tools still struggle a bit to compete with expert developers – e.g. developers with 10+ years experience. But in many ways they're already better at a large chunk of their job because an expert developer will still spend a good amount of their time on minor refactoring which AI can do much faster. But for now AI + an expert is still the optimal productivity mix for performance-optimised companies.
I suspect in a 2-3 years 95+% of expert development work will be doable with AI agents. Obviously there will be niches where you need will need human picking up the last bits, but the percentage for which human experts are required will exponentially decrease with every passing year.
I guess the good news is like how outsourcing massively reduced the cost of physical goods, AI will massively reduce the cost of digital goods. There will of course be those who complain that "foreigners are taking our jobs!" or "robots are taking our jobs!" but there will probably still be service jobs like waiting tables or serving coffee that AI can't replace.
Nothing that we have now is AI in its true form, its just super sophisticated search and automation, for which you still need human direction.
There is also nothing that even looks like AI on the horizon at this point.
I'd say the notion of what programming is will shift (again). Programmers might use AI helpers, or embed AI functions in you products like we are using IDEs and libraries since a while.
I'd also say that many established programmer roles today are already beginning to be disrupted.
But maybe I am wrong. Deploying and operating a web-based product in a secure manner is often not as easy as it looks like.
Also, as of now, there needs to be someone scripting the model training, and all the "logistics" around it.
For my whole life, the prediction that (ro)bots will do all the work has unfortunately never materialized. To me it looks as likely as having economically feasible fusion energy in the grid within the next 30 years ;-)