They dont need to automate all office jobs to turn a profit. Theres about 18 million office workers in the US and they make an average of around 50k per year. Thats about 900 billion per year, and thats just the US without taxes, office space, utilities and other overhead. If AI could increase average office worker productivity by 50% and OpenAI could capture most of that they could pay back their investment within just a few years and become the most profitable company the world has ever seen. The problem, however, is not just technical, its is going to be capturing that economic windfall, and keeping the savings from being passed on to clients and consumers. Political backing from a reckless and authoritarian autocrat might go a long way towards securing that goal.
The trade-offs and fine-print and all the hidden steps and caveats to everything - it’s like at least 10 years too early, but the unfathomable hype frenzy means we’re going to get stick with something. A generalizing summary bot is - OK.
If they can ever get it to not have to be double-checked every time it’d be more useful. But the fact is, AI simply doesn’t understand what words mean. Math is good, code is a sort-of in-between. Everything else is a crapshoot.
They dont need to automate all office jobs to turn a profit. Theres about 18 million office workers in the US and they make an average of around 50k per year. Thats about 900 billion per year, and thats just the US without taxes, office space, utilities and other overhead. If AI could increase average office worker productivity by 50% and OpenAI could capture most of that they could pay back their investment within just a few years and become the most profitable company the world has ever seen. The problem, however, is not just technical, its is going to be capturing that economic windfall, and keeping the savings from being passed on to clients and consumers. Political backing from a reckless and authoritarian autocrat might go a long way towards securing that goal.
I mean, that cart is miles ahead of the horse.
AI agents that can replace humans are miles away, but AI that can improve human output isnt.
Mmmmmmmm . . . Sort of kind of, to some extent.
The trade-offs and fine-print and all the hidden steps and caveats to everything - it’s like at least 10 years too early, but the unfathomable hype frenzy means we’re going to get stick with something. A generalizing summary bot is - OK.
If they can ever get it to not have to be double-checked every time it’d be more useful. But the fact is, AI simply doesn’t understand what words mean. Math is good, code is a sort-of in-between. Everything else is a crapshoot.