I think you're underselling the investment not just of capital but also just of inertia it's gonna take to adopt this stuff at a larger, corporate level. It could be rapid changes... we're also dealing with a brand new thing that corporations weren't planning on doing this year. RPA as an accelerator has by and large been a failure. This is like RPA 8.0, but it's still got a lot of limitations and companies won't let the unknown of those limitations impact their business. We're going to see a lot of internal POCs, a lot of dipping the toe in the water... not a lot that is customer-facing for quite awhile and not a lot that impact critical internal systems either. It'll start by just being an accelerator. Help your dev write a line of code that's stumping them. An optional content management/search functionality to help you find that one document on your company's intranet that you just can't find. Plug-ins to Office 365 to accelerate tasks that you're still very much in control of. There won't be this big cutover overnight.
I think there will be lots of areas where it will be adopted very rapidly. Look at paralegals, as a possible example. Right now they spend a lot of time looking up information in legal texts and then writing summaries for the lawyers that ask for them. Maybe a law firm has one paralegal per lawyer, so a large law firm needs a dozen or so of them, maybe? Well, they've already shown that these GPTs and LLMs can pass a bar exam 90% of the time. And it takes seconds to scan the legal texts and create a written summary of information. So now a firm with a dozen paralegals can have two or none depending on what the lawyers are willing to do and pay for.
Same with coding. These GPTs and LLMs will do a lot of the grunt work that is being done by well paid programers. And while I have seen no data on this, I bet they are much better at error testing than humans as well. Plus, if you don't need nearly as many people, you don't need nearly as many janitorial staff, human resources, etc. The cascade of the loss of the white collar workers will hit supporting blue collar work as well.