I'll look at your face when you lose your job because of some innovation.
"First they came for the socialists, and I did not speak out—
Because I was not a socialist.
Then they came for the trade unionists, and I did not speak out—
Because I was not a trade unionist.
Then they came for the Jews, and I did not speak out—
Because I was not a Jew.
Then they came for me—and there was no one left to speak for me."
The one counter point to that is each of your other examples (besides maybe sharp stones) still leads to a fairly comparable number of jobs and as such a fairly comparable number of people supported by said jobs. The weavers were replaced by people who handled and maintained the looms, similarly horse breeders were replaced by mechanics and factory workers. AI has a much harsher drop off due to the sheer number of potential applications compared to the number of people required to maintain them. We're already in a bit of a dicey situation for low end workers right now given the weakening of unions over the last 50 years, a sudden influx of unemployment due to a mass automation of admin or creative positions could lead to some very troubled times indeed.
My point was that it will be, and already are, used as a
tool. And this moves discussion from 'thats a pretty handy tool' to sociology.
It cant replace everyone, purely from the fact that there are should be someone with experience who will check result of neural net is actually what it supposed to do or no. And somebody should be able to maintain/generate requests for neural nets.
My friendo who draws textures, model models and similar shit uses locally trained stable diffusion, IIRC, to generate ideas for models/textures/etc. He even made a photosession of my basement for a 'cool abandoned factory style', because there were no model with similarly trained dataset.
I personally used/trained one to generate some cool ideas for a doom (1/2) maps but it semi abandoned now because my time now is occupied with how to make a compiler/interpreter with dedicated VM.
And most complains about 'neural nets steals our job' is from kind of artists that draws tobs graphomany kind of shit.
Actual jobs that were 'optimized' by neural nets is that kind of jobs that should be illegal by default, like accounting from 8 to 8/9/10, 6/1 on minimal wage. Im aware the people who picked such jobs done it not because they live life on hardcore, unlike of those pesky casuals, but it WILL improve overall quality of job market/laws in perspective. Dont know is it will be short or long perspective, but there will be changes in this field. But this is a first world problem.