So RTO is just an excuse for layoffs, I guess?
They're copying a technique from the minimum wage employer playbook, where you would lay people off, replace them with whoever (or no one at all) and rehire your original staff once they softened up while firing those new hires who proved problematic.
Of course companies don't do it in concert, so weirdly it's easy to both get hired and fired in this space, but net effect is that people keep getting paid minimum wage.
And therein lies the crux of the issue - our employers are pretty well synchronized in their layoffs so they're not worried that their projects will stall for a few months, as everyone else's will as well.
In a competitive market, there would be defection from any layoff cartels, to gain market share.
If the employer has a monopoly, they are mostly competing with themself.
Here it seems that the moment LLMs hit the market in earnest, every C-level started asking themselves if they have the manpower to compete in this space and ultimately reached the conclusion that no, they don't, so they started doing layoffs, (correctly) assuming that everyone would have the same strategy.
We will all be rehired eventually, but the perception of our value vs that of AI is lower and that is reinforced by the number of people currently looking for a role.
In a sense the press was right - developers were replaced with AI - at least temporarily.
Funny thing is this happened to a project I was in back in 2017. The company figured that instead of building a huge system using a team of 40+ people they'll... use Watson.
That could mean 2 more years of LLM CapEx running air cover for engineer labor OpEx tax policy, which ironically puts humans and servers/GPUs/robots/AI on a comparable depreciation schedule. Meanwhile, an LLM subscription would be immediately deductible, while the developer's salary would require 5 years, https://news.ycombinator.com/item?id=34627712