I guess I just feel completely different about those conditional probabilities.
Unless we hit another AI winter the profit and national security incentives just snowball right past almost all of those. Regulation? “Severe depression”
I admit that thr loss of taiwan does innfact set back chip manufactyre by a decade or more regardless of resoyrces thrown at it but every other case just seems way off (because of the incentive structure)
So we’re what , 3 months post chatgpt and customer service and drive throughs are solved or about to be solved? , so lets call that the lowrst hanging fruit. So just some quick back of the napkin google fu , the customer service by itself is a 30 billion dollar industry just in the US.
And how much more does the math break down if say , we have an AGI that can do construction work (embodied in a robot) at say 90% human efficiency for...27 dollars an hour?
In my mind every human task fully (or fully enough) automated snowballs the economic incentive and pushes more resources and man hours into solving problems with material science and things like...idk piston designs or multifunctionality or whatever.
I admit I’m impressed by the collected wisdom and apparent track records of these authors but it seems like its missing the key drivers for further improvement in the analysis.
Like would the authors have put the concept of a smartphone at 1% by 2020 if asked in 2001 based on some abnormally high conditionals about seemingly rational but actually totally orthagonal concern based on how well palm pilots did?
I also dont see how the semi conductor fab bottleneck is such a thing? , 21 million users of openai costs 700k a day to run.
So taking some liberties here but thats 30 bucks a person (so a loss with their current model but thats not my point)
If some forthcoming iteration with better cognitive architecture etc costs about that then we have , 1.25$ per hour to replace a human “thinking” job.
Im having trouble seeing how we don’t rapidly advance robotics and chip manufacture and mining and energy production etc when we stumble into a world where thats the only bottleneck standing in our way to 100% replacemwnt of all useful human labor.
Again , you got the checkout clerks at grocery stores last decade. 3 months in and the entire customer service industry is on its knees. Even if you only get 95% as good as a human and have to sort of take things one at a time to start with , all that excess productivity and profit then chases the next thing. It snowballs from here.
I guess I just feel completely different about those conditional probabilities.
Unless we hit another AI winter the profit and national security incentives just snowball right past almost all of those. Regulation? “Severe depression”
I admit that thr loss of taiwan does innfact set back chip manufactyre by a decade or more regardless of resoyrces thrown at it but every other case just seems way off (because of the incentive structure)
So we’re what , 3 months post chatgpt and customer service and drive throughs are solved or about to be solved? , so lets call that the lowrst hanging fruit. So just some quick back of the napkin google fu , the customer service by itself is a 30 billion dollar industry just in the US.
And how much more does the math break down if say , we have an AGI that can do construction work (embodied in a robot) at say 90% human efficiency for...27 dollars an hour?
In my mind every human task fully (or fully enough) automated snowballs the economic incentive and pushes more resources and man hours into solving problems with material science and things like...idk piston designs or multifunctionality or whatever.
I admit I’m impressed by the collected wisdom and apparent track records of these authors but it seems like its missing the key drivers for further improvement in the analysis.
Like would the authors have put the concept of a smartphone at 1% by 2020 if asked in 2001 based on some abnormally high conditionals about seemingly rational but actually totally orthagonal concern based on how well palm pilots did?
I also dont see how the semi conductor fab bottleneck is such a thing? , 21 million users of openai costs 700k a day to run.
So taking some liberties here but thats 30 bucks a person (so a loss with their current model but thats not my point)
If some forthcoming iteration with better cognitive architecture etc costs about that then we have , 1.25$ per hour to replace a human “thinking” job.
Im having trouble seeing how we don’t rapidly advance robotics and chip manufacture and mining and energy production etc when we stumble into a world where thats the only bottleneck standing in our way to 100% replacemwnt of all useful human labor.
Again , you got the checkout clerks at grocery stores last decade. 3 months in and the entire customer service industry is on its knees. Even if you only get 95% as good as a human and have to sort of take things one at a time to start with , all that excess productivity and profit then chases the next thing. It snowballs from here.
I agree the point about freeing up resources and shifting incentives as we make progress is very important.
Also, if your 21 million open AI users for $700k/day numbers are right, that’s $1/user/month, not $30/user/day. Unless I’m just misreading this.