Someone with a large following on X wrote the following:
I do believe AI is going to transform the world. And this transformation will be much more dramatic and much faster than many people think. But such statements need to be taken with a grain of salt.
In 2011, Google DeepMind co-founder Shane Legg gave me the following probabilities for the development of artificial general intelligence (AGI):
10% / 50% / 90% = 2018, 2028, 2050.
I still consider the estimates for 2028 and 2050 to be realistic. This means that I continue to believe there's at least a 50% chance that AGI won't be achieved in the next 3–4 years.
However, even if we do reach AGI by then, it doesn’t mean we will suddenly be living in a utopia. Initially, such a model would be very inefficient and would only run on a supercomputer. It could take another 2–3 years of algorithmic progress and advances in computing power before it would be cost-effective to have this AI perform tasks instead of humans.
Moreover, several other hurdles could slow down the process. On the one hand, governments might intervene directly and halt development out of fear, but this is unlikely. By then, we would likely already be in an existential arms race. On the other hand, many existing laws prohibit the automation of certain processes and professions. It will also take time for the AI to advance from human-level intelligence to superhuman levels, although this latter transition could happen relatively quickly.
The most important caveat, however, is that many breakthroughs require slow, experimental work that cannot simply be simulated. It is highly unlikely that a superintelligence will simply "dream up" a cure for aging without conducting some empirical experiments.
All in all, the utopia described in this post may still be a long way off—probably at least 20 years.
Note: A dystopian outcome, such as extinction, could occur much more quickly.
I’m being 100% serious:
Your only goal the next 3 years is to not die
We are about to enter the greatest age in human history
AGI. Going to mars. Reverse aging. Humanoid robots. Self driving cars.
...
There is a decent chance in the next 3 years we figure out how to reverse aging and nobody dies anymore
I do believe AI is going to transform the world. And this transformation will be much more dramatic and much faster than many people think. But such statements need to be taken with a grain of salt.
In 2011, Google DeepMind co-founder Shane Legg gave me the following probabilities for the development of artificial general intelligence (AGI):
10% / 50% / 90% = 2018, 2028, 2050.
I still consider the estimates for 2028 and 2050 to be realistic. This means that I continue to believe there's at least a 50% chance that AGI won't be achieved in the next 3–4 years.
However, even if we do reach AGI by then, it doesn’t mean we will suddenly be living in a utopia. Initially, such a model would be very inefficient and would only run on a supercomputer. It could take another 2–3 years of algorithmic progress and advances in computing power before it would be cost-effective to have this AI perform tasks instead of humans.
Moreover, several other hurdles could slow down the process. On the one hand, governments might intervene directly and halt development out of fear, but this is unlikely. By then, we would likely already be in an existential arms race. On the other hand, many existing laws prohibit the automation of certain processes and professions. It will also take time for the AI to advance from human-level intelligence to superhuman levels, although this latter transition could happen relatively quickly.
The most important caveat, however, is that many breakthroughs require slow, experimental work that cannot simply be simulated. It is highly unlikely that a superintelligence will simply "dream up" a cure for aging without conducting some empirical experiments.
All in all, the utopia described in this post may still be a long way off—probably at least 20 years.
Note: A dystopian outcome, such as extinction, could occur much more quickly.