I'm a unionized software developer, and have been part of two separate organizing campaigns at two different employers.
Happy to answer questions about how it all works - there are only about ~6000 dev/product/design/qa employees in the U.S., so there is very little first hand experience to go around.
Among other things that my coworkers and I secured by bargaining collectively are:
- Guaranteed remote work.
- Guaranteed forty hour work week, with compensatory time within two weeks if management asks us to work overtime.
- Not on call 24/7/365. Everyone is in a defined rotation, and you are guaranteed three hours off just for being on call for a week, and a whole day if you receive a call. I believe that partly because of this policy, after hours incidents are extremely infrequent now.
- Guaranteed floor on wage increases every year.
- Just cause for discipline. People still get terminated if they don't do the work, but if you want it (and almost everyone does) there is an elected coworker in your corner to guide you through a PIP and make sure standards are enforced evenly. You can't suddenly be terminated for no reason.
- Extra time off each year through self-directed times between sprints and quarterly increments.
- Right to review all code done by outside contractors if you're going to have to maintain it in the future.
Former BLS officials still believe the numbers are reliable, because the methodology is public and there are lots of ways for well-financed actors (e.g. wall street traders) to spot manipulation.
To go by the numbers, it's pretty good. Not as good as the truly anomalous post-pandemic peak of hiring. But still good compared to almost every other occupational group.
I think many of these problems still arise if inference is effectively free in monetary terms to the end user. In many economic processes, time to getting the final and correct answer is the major driver of profitability.
I'm no expert in European religious history, but it is an interest of mine and I've read a fair bit more than the typical American about it.
I am certain that every church father and major theologian from antiquity to the early modern period would have condemned without reservation the very idea of building a machine to answer theological questions. Possible exceptions, like Ramon Llull, prove the rule.
Given the prevalence of stories about Djinn and Golems in Islamic and Jewish literature, it feels safe to say they would have looked askance at this idea as well.
"There’s a lot of boring work AI can automate with minimal risk. There’s also the potential to decrease risk with AI too, including ensembles of different AIs modals and AI + human."
I think the trouble, economically speaking, is that while it will be possible from a purely technical standpoint to unbundle a job performed by a human into separate tasks, many of which can be "done" by agents, the new process will not present a cost savings overall once the entire lifecycle of the task is taken into account. The economist David Autor has written about these challenges extensively, and his theory accords with my experiences.
Why prove the Pythagorean theorem rather than just prove 3^2 + 4^2 = 5^2?
For any practical application, you are only interested in finite set of concrete identities, so anything beyond that is surplus to requirements, surely?
I am very sympathetic to your situation, and fear something similar happening to me. But the article is addressed to investors, who are interested in aggregate labor demand declines.
"A chart showing “total number of jobs” is not meaningful."
It is meaningful in answering the question being asked, which is whether the hype around labor displacement, which has been growing for nearly five years now, is actually occurring in a way that would justify some of the higher valuations for AI firms.
The "AI jobs crisis" is generally understood to mean a sustained downturn in demand for all labor due to AI substituting for labor across a huge range of tasks.
"But then people who work in actual tech companies come in and explicitly say they are not hiring any juniors anymore specifically because AI is good enough to do most of what juniors do, and that senior engineers can now write 3x as much code, etc."
If you want an anecdote: the media company I work for just started hiring interns and juniors in software career tracks again after a lengthy hiatus.
I'm trying to envision the attacks and countermeasures for this approach. I suppose what I would do to fake this is write a program that takes text and the desired output histogram of edit timings and randomly introduces jitter into the typing. Then, I'd imagine you'd develop more sophisticated statistical tests, and so on. My intuition is that eventually there would be no statistical test that reliably discriminates real human typing.
http://bjornwestergard.com