10 comments

[ 3.0 ms ] story [ 24.8 ms ] thread
How is causality figured out here? How do we know it's not just a "standard" market downturn/downsizing to compensate for previous overhiring etc.?
How many of these jobs will stay gone? I feel like I've seen this over and over where a new advancement causes job loss then management realizes it wasn't all it was cracked up to be and need to hire again.
It's not replacing jobs. It's deferring their creation until the executives finally realize they fucked up and you can't replace a human with a glorified search engine that spews complete bullshit 30 percent of the time.
I had a rather unsettling thought occur to me the other day.

There are many people in this world, to whom truth and reality are fungible, malleable, unimportant things in the face of their agendas. These same people are overrepresented at the highest rungs of the corporate ladder, in the highest echelons of the halls of power in our world.

For them, right 70% of the time is more than good enough, because that which is real has always taken a back seat to that which is expedient.

So what if the Plagiarism Engine spews Confidently Stated Bullshit 30% of the time? They themselves do that at least as often, and for far more than the cost of what LLM's demand per call. Besides, they're so used to "perception=reality" that they figure for the 30% Confidently Stated Bullshit, they can just paper over it, bully, disconcert and repeat until reality matches their expedient bullshit.

The sad thing is, so much of our economy and society is based on bullshit and scams now, they may not be wrong. In fact, they're probably right, for at least a significant percentage of the populace who believes chemtrails are making the frogs gay etc.

I suspect some very hard lessons will need to be learned from disciplines where rigor is required 100% of the time, else lives are lost, and lives will be lost, because the business idiots will continue to shoehorn this garbage in everywhere they possibly can, to sustain the hype cycle bullshit they're all cashing in on. Only once many lives have been lost, and the link conclusively drawn to the garbage spewed by these statistical wordcloud predictive autocorrect machines, then we might see a modicum of forceful pushback. But I ain't holding my breath

"Over the past two years, there has been a 400% increase in employers using AI in job descriptions, the firm found."

How is that replacing?

AI isn't about jobs and efficiency, it's about having a stronger position over labor. AI is useful, in most cases this makes existing labor better. The number of jobs that can be actually automated by it is much lower than people percieve. But the narrative is what matters. The temporary displacement and uncertainty of labor is what matters. It creates a weaker position for labor.
I tried ChatGPT 5 and Claude for some medium, non-trivial coding tasks in Rust, and most of the time the code does not even compile. It may work better for other programming languages. However, this make me believe that unless you want basic small functions (like convert integer to string) done by AI, betting on it to replace SDEs is still risky, as of today. On the other hand, AI tech is progressing quickly.