The internet revolutionized many industries and fundamentally uprooted processes that we all depend on a daily basis. Chief among these is distribution. Distribution of goods and services, distribution of communication, and distribution of information. While this certainly changed the way we all do things, it replaced the existing system with a new one. This pertains to commerce, education, professional development, and social interaction. I have openly criticized the way social interaction has changed in the internet age, but the point is that it has been changed, not that the changes have been wholly positive. The open question remains: will artificial intelligence do the same or will there be no human system that can replace what's taken?
We don't like non-answers to existential questions. To be in the dark on such a societally destructive potentiality is unacceptable, but… it's where we are, if we're honest. Are we to become a worldwide commune that produces artisan goods in our infinite leisure time or a UBI hellscape where people lose their ambition and turn to stimulation of any kind to try to feel like they're living an actual life? No one knows and those who claim to know know even less.
One thing that people in the tech world never seem to add to the equation is that the rate of adoption never matches the rate of innovation. Blue chip companies aren't going to convert their business models overnight even if there is some AI that can produce as much as a human. Plus, with the possibility of AI/Robot Taxes being levied on corporations as this technology slowly becomes more utilized, this is a much more nuanced picture than either the utopian or dystopian visions provided.
Also worth noting: an AI has not been created that can direct itself, nor has an AI been created that has legitimately innovated anything without human direction. So are LLMs more like a language-calculator than a God? Only time will reveal this. An unforeseen boundary in the development of these systems could crop up at any moment, in fact, they may even have revealed themselves already.
AIs struggle with incredible amounts of characters being thrown at them all at once; any programmer will tell you this and there have been studies as well. How, without direction, will an AI manage an entire code base if their response reduces in quality when provided a fraction of that code? When AI is finally trained on all available human data and the corpus of human achievement has been eaten by these machines, won't the synthetic data generated by these machines mirror the data that the LLMs were trained on? Again, these aren't innovative machines, maybe we'll have something like that in the future, but LLMs predict characters based on user input, context, and training data through features and weights. How can we expect to surpass human intelligence if that intelligence hasn't been recorded?
There's no definitive answer I can give to the incredible variety of questions that this technology inspires, but I can venture a guess based on what I've seen in my limited 27 years: we'll be fine. Probably. Maybe.