Skip to main content


N. E. Felibata 👽 reshared this.

That's fair, and I might be misreading it. But I get a very strong "industrial = slop" vibe from it, e.g. :

industrialisation of printing processes led to paperback genre fiction
industrialisation of agriculture led to ultraprocessed junk food
industrialisation of digital image sensors led to user-generated video


He seems to pretty much always describe anything produced "industrially" as of lower quality than that made by hand.

Isn't it?

I.e. typically industrialisation leads to many compromises required for cost savings, manufacturability, and of course creation of a more "generic" product.

Of course there are many counter-examples where non-industrial processes have lower quality, and many of those no doubt apply to software creation. In the end though I don't think they outweigh the losses incurred, at least not for the average person.

Indeed as I read the whole article carefully I find the author making exactly the same claims, e.g.:

Industrial systems reliably create economic pressure toward excess, low quality goods.
Ultimately I still believe (or at least hope) that LLM/GPT systems will fall out of favour because of their very limited ability to produce reliable quality output. They are a dead end in AI progress.

It's fun to watch archaeology shows which aren't digging up the homes of the rich and famous. Very quickly it becomes apparent that the average product quality was a lot lower than even IKEA's basic range. The extremes will be different, and debatable, but aren't really relevant when considering the overall effect.

I think LLMs are certainly a dead end in terms of intelligence/consciousness research, although it's fascinating to me just how much they can mimic of true human intelligence. I reserve judgement in whether they're a dead end overall, although I'd be surprised if they aren't an integral part of most AI services for the next decade at least. What they do well they can do very well.

He seems to pretty much always describe anything produced "industrially" as of lower quality than that made by hand.


I guess I interpreted it differently -- that industrialization doesn't make high quality go away -- he says, for example, "There is a thriving and growing demand for healthy, sustainable production of foodstuffs" -- but adds a new "low quality" dimension. He seems to be making the point that "industrialization" floods the market with low quality goods, and low quality in a way not possible before, e.g. ultraprocessed foods were simply not possible before industrialization. We didn't get an exponential increase in health, we got a juxtaposition of starvation and obesity.

On the contrary - an exponential increase in health is exactly what we got ! Sanitation, vaccines, reliable crops, controlled temperatures, anaesthetics, transplants... It's not industrialisation (manufacture at scale) that's the problem at all - it's the business model.
⇧