alt Hacker NewsUSA had a nearly constant per person economic growth rate of 2%/year in the last ~150 years, perhaps going as far back as the beginning of industrial revolution.
Extrapolating such curves into the future suggests the current AI revolution is simply the last and latest node in a string of revolutions and it's nothing special.
If you think about it, having the world's all information at your fingertips (google), and in your pocket (iphone) might have been equally revolutionary. And before that came TV, radio, car, train, boat, plane, electricity, gas engine, steam engine etc as revolutions.
There's nothing that suggests the economic output per person is accelerating beyond the historical 2%/year. What could be reasons? Perhaps limited electricity, compute, AI model quality, computer speed etc.
So, the more analytical side of me thinks what we're experiencing is nothing extraordinary. It's just another revolution in a string of many :)
Obviously, my other, the more human side gets scared and feels afraid about the meaning of life, and humanity's place in it.
Replies
Yes, for information and reference we already had Wikipedia and billions of web pages indexed in Google, searchable by keyword. For questions we had reddit, StackOverflow and forums. For chatting we had social networks, chatting with real humans. For image we had only search, but within billions of images. Faster than gen AI, and made by humans. For code we had hundreds of thousands of repos.
We already had the genAI goodies for 2 decades. It's not going to be such a shocking change.
Statistical next-token predictors that aren’t even correct some of the time, and are currently crafted to pass tests, isn’t what I would consider revolutionary.
They’re neat tools. They help some people (a much, much smaller group of people than most think) be a bit more productive.
If LLMs are considered revolutionary, we are stagnating.