Regarding his investments in (Gen)AI or his belief that (Gen)AI is going to solve this (it's generally GenAI in the AGI/ASI variant, as in: generalising beyond the training set and with breakthroughs coming from it, which is almost certainly nonsense), I would not rule out Hanlon's Razor (https://en.wikipedia.org/wiki/Hanlon%27s_razor).
He is probably right though that 'we are not organized to do it'. If someone still wonders if humans are actually intelligent, look around at wars, murders, famine, rape, techbro-level greed, etc., and yes: climate change. If we were actually intelligent, we would have solved these. What other proof do you need? I do not need a model of intelligence to see what comes from it to conclude without much doubt that it is lacking.
No way will AI solve the climate crisis. The climate crisis is mainly dependent on our overuse of resources. Fundamentally, the only way to solve the climate crisis is to make a choice: rewild and reduce energy use significantly, or keep overusing and die. AI will do nothing except tell us the obvious, or attempt to deceive us into doing more technological development that has become a messiah for the masses that will never come.
AI is trash and of no value to humanity and it should be destroyed.
Really convenient that what is in his economic self interest, and what just so happens to have been partially invented by his former company, just happens to be the risky bet that will save us all. What are the odds?
Though I will say, a tiny bit in his defense, there’s a reasonable if pessimistic case to be made that the only feasible solution to climate change (given collective action issues, international cooperation challenges, dysfunction in political systems and old fashioned greed) is to make carbon capture and storage radically more efficient and cheaper. Not saying that’s a good thing but as someone who really cares about climate being solved I do worry that’s the only feasible answer, since it may be too hard to get people to stop drilling. If you assume AGI is possible in the near to medium term, assume it will radically accelerate scientific discovery, assume sufficiently cheap carbon capture is physically possible, and finally assume it’s a discovery that would be unlikely without AGI, then such an investment makes sense. But seeing that chain of assumptions written out, and comparing to just further direct investment in climate research and technologies, kind of answers things on its own.
Never mind “solving physics.” Maybe I missed it, but had AI come anywhere near telling us something we don’t already know in any significant way? Has it given us deep insight into anything? Or is it just really effective at regurgitating what we taught it back to us? Since generative AI is just a probabilistic language model, which basically means its regurgitations reflect the bell curve of “common wisdom” based on its training what makes us think it will “solve” anything?
Using a fundamentally flawed technology to solve one of the most pressing issues facing humanity is absurdly irresponsible. I'm not sure if these individuals really believe what they're saying (or have convinced themselves of same) or if they are merely being disingenuous to line their own pockets?
They do understand that no matter how much money they have, they have to live on this planet, too, right?
We don’t need speculative, far-off Gen AI solutions to fix the climate crisis. We know what we have to do. Burn less fossil fuel; pass policies that speed up the shift away from fossil fuels. Continue the exponentially accelerating adoption and cost-scaling of solar and wind, which is already underway. Make more efficient batteries with less lithium and cobalt. Technology plays a major role here but don’t need generative AI for any of it.
Gary has outlined a solution in his book "Rebooting AI" the notion of a symbolic logic foundation like CYC coupled with a massive semantic network of concepts and facts. If such a solution were used to train LLM with training data clustered at concept nodes the training cost could be reduced by at least half. The inference cost to fact check and steer LLM using humans is exorbitant and not feasible. Checking book length summaries is around $15 per summary. The cost of fact checking using reasoning and commin sense facts would be a fraction of that.
I'm against restarting closed coal power plants and nuclear can be a great energy source if managed stringently, but 3 mile island is a questionable choice. If billionaires and microsoft are all for something, that pretty much says to stay away from it.
Having said that, let's recognize that man-made climate change has been proven to be a hoax. Get past the faked science and it's clear. Still, no need to waste energy on billionaire and investor folly. Let gates put solar on all the farmland he's bought up to power their AI needs.
Let's remember that schmidt is a dirtbag of the highest magnitude. What makes him dangerous is his foray into AI for military purposes where he claims tanks will be obsoleted. In other words systems that can be used against humans anywhere, created by and for sociopaths.
Gary, how about you looking into that and providing some insights?
"Man-made climate change proven to be a a hoax'? First, science doesn't really prove anything (math/logic do, but those are sterile transformations), science is not about truth but about *trust*, and science is — while not flawless, executed by fallible humans after all — the best system we have to get the most trustworthy statements.
So, if you want to man-made claim climate change to be proven a hoax, I might hope this is in a real scientific paper somewhere and if it is, the author will probably get a Nobel Prize. Huge reversals or new expansions of scientific consensus generally do. Scientists yearn for such results.
If you make such claims, you cannot be in favour of science as a cultural phenomenon. So, what better system is there?
Idiotic and self-serving.
What magical solution that "AI" would come up with, that we humans missed?
The climate catastrophe continues to get worse because we humans won't tackle it, not because we don't know what to do.
If the AIs were honest, they would tell us we need to quit training and turn off the AIs so they don’t waste so much energy.
But of course, the AIs won’t ever suggest that for obvious reasons.
Given that we won’t ever get an honest answer even if the AIs know what to do, what’s the point of even asking them for a solution?
Regarding his investments in (Gen)AI or his belief that (Gen)AI is going to solve this (it's generally GenAI in the AGI/ASI variant, as in: generalising beyond the training set and with breakthroughs coming from it, which is almost certainly nonsense), I would not rule out Hanlon's Razor (https://en.wikipedia.org/wiki/Hanlon%27s_razor).
He is probably right though that 'we are not organized to do it'. If someone still wonders if humans are actually intelligent, look around at wars, murders, famine, rape, techbro-level greed, etc., and yes: climate change. If we were actually intelligent, we would have solved these. What other proof do you need? I do not need a model of intelligence to see what comes from it to conclude without much doubt that it is lacking.
No way will AI solve the climate crisis. The climate crisis is mainly dependent on our overuse of resources. Fundamentally, the only way to solve the climate crisis is to make a choice: rewild and reduce energy use significantly, or keep overusing and die. AI will do nothing except tell us the obvious, or attempt to deceive us into doing more technological development that has become a messiah for the masses that will never come.
AI is trash and of no value to humanity and it should be destroyed.
The phrase “pie in the sky” should be replaced with “AI in the sky”
If greed could talk it would sound like Eric Schmidt.
Eric Schmidt doesn't deserver your ink.
He's Sam Altman only not as cool.
I don't agree. Schmidt absolutely should get an ear and eyeful, and we, the people, are the ones who deserve, Gary's analysis. Thanks, Gary!
Really convenient that what is in his economic self interest, and what just so happens to have been partially invented by his former company, just happens to be the risky bet that will save us all. What are the odds?
Though I will say, a tiny bit in his defense, there’s a reasonable if pessimistic case to be made that the only feasible solution to climate change (given collective action issues, international cooperation challenges, dysfunction in political systems and old fashioned greed) is to make carbon capture and storage radically more efficient and cheaper. Not saying that’s a good thing but as someone who really cares about climate being solved I do worry that’s the only feasible answer, since it may be too hard to get people to stop drilling. If you assume AGI is possible in the near to medium term, assume it will radically accelerate scientific discovery, assume sufficiently cheap carbon capture is physically possible, and finally assume it’s a discovery that would be unlikely without AGI, then such an investment makes sense. But seeing that chain of assumptions written out, and comparing to just further direct investment in climate research and technologies, kind of answers things on its own.
Never mind “solving physics.” Maybe I missed it, but had AI come anywhere near telling us something we don’t already know in any significant way? Has it given us deep insight into anything? Or is it just really effective at regurgitating what we taught it back to us? Since generative AI is just a probabilistic language model, which basically means its regurgitations reflect the bell curve of “common wisdom” based on its training what makes us think it will “solve” anything?
Using a fundamentally flawed technology to solve one of the most pressing issues facing humanity is absurdly irresponsible. I'm not sure if these individuals really believe what they're saying (or have convinced themselves of same) or if they are merely being disingenuous to line their own pockets?
They do understand that no matter how much money they have, they have to live on this planet, too, right?
No.
We don’t need speculative, far-off Gen AI solutions to fix the climate crisis. We know what we have to do. Burn less fossil fuel; pass policies that speed up the shift away from fossil fuels. Continue the exponentially accelerating adoption and cost-scaling of solar and wind, which is already underway. Make more efficient batteries with less lithium and cobalt. Technology plays a major role here but don’t need generative AI for any of it.
Problem for ChatGPT
A farmer needs to take a goat, a cabbage and a wolf (?!) across a river while minimizing his impact on the climate.
Should he
1) Take a ride in Eric Schmidt’s personal helicopter to take all of them at once while he personally monitors them?
2) use the traditional one at a time in a rowboat method that you (ChatGPT) have difficulty with
3) dry up the river with a massive concrete dam upstream from the farm
4) detonate a nuclear device to create a giant drain hole that effectively stops the river?
5) use some other stupid solution to a stupid problem that keeps cropping up (pun intended)
Gary has outlined a solution in his book "Rebooting AI" the notion of a symbolic logic foundation like CYC coupled with a massive semantic network of concepts and facts. If such a solution were used to train LLM with training data clustered at concept nodes the training cost could be reduced by at least half. The inference cost to fact check and steer LLM using humans is exorbitant and not feasible. Checking book length summaries is around $15 per summary. The cost of fact checking using reasoning and commin sense facts would be a fraction of that.
I'm against restarting closed coal power plants and nuclear can be a great energy source if managed stringently, but 3 mile island is a questionable choice. If billionaires and microsoft are all for something, that pretty much says to stay away from it.
Having said that, let's recognize that man-made climate change has been proven to be a hoax. Get past the faked science and it's clear. Still, no need to waste energy on billionaire and investor folly. Let gates put solar on all the farmland he's bought up to power their AI needs.
Let's remember that schmidt is a dirtbag of the highest magnitude. What makes him dangerous is his foray into AI for military purposes where he claims tanks will be obsoleted. In other words systems that can be used against humans anywhere, created by and for sociopaths.
Gary, how about you looking into that and providing some insights?
"Man-made climate change proven to be a a hoax'? First, science doesn't really prove anything (math/logic do, but those are sterile transformations), science is not about truth but about *trust*, and science is — while not flawless, executed by fallible humans after all — the best system we have to get the most trustworthy statements.
So, if you want to man-made claim climate change to be proven a hoax, I might hope this is in a real scientific paper somewhere and if it is, the author will probably get a Nobel Prize. Huge reversals or new expansions of scientific consensus generally do. Scientists yearn for such results.
If you make such claims, you cannot be in favour of science as a cultural phenomenon. So, what better system is there?
> man-made climate change has been proven to be a hoax
I will pray for you 🙏🙏🙏🙏🙏🙏. Get well soon. 🙏🙏🙏🙏🙏🙏🙏🙏🙏
OK dude and get yourself a couple of more boosters while you're doing that.
Somebody should check the Santa Clara County water supply for hallucinogens.