If AI reaches the point where it can do the cognitive work humans do, robots will proliferate. Standard data on US industry implies a fully automated economy could double its output roughly every year
Your article usefully takes seriously what many economic forecasts avoid: if cognitive work and physical work become automated, production capacity could expand at an extraordinary rate.
MAny (me) would argue the central question is whether an industrial explosion is a sensible objective on a finite planet or indeed a threat multiplier to be avoided
Your article appears to treat the economy largely as a self-reinforcing system of industrial inputs and outputs. Yet the economy is embedded in the Earth system. Energy, materials, land, water, ecological stability and waste-absorption capacity are not merely inputs that can be scaled indefinitely. They are physical conditions, bounded in quantity and constrained by time. Annual doubling of material output is not just a remarkable growth rate; it rapidly becomes an ecological impossibility. Remember the duckweed on a pond that doubles each day?
There is also a question of humans and democracy. If AGI and robotics remove labour as the limiting factor, who determines what is produced, for whom, and to what end? Cheap automated production does not itself distribute shelter, care, security, dignity or political agency. Without deliberate governance, it most likely will concentrate ownership, productive power and the capacity to shape public reality.
Alignment must extend beyond the behaviour of the machine and the few who benifit. A system can be perfectly aligned to an objective and still be catastrophically misaligned with human wellbeing, democratic legitimacy and Earth-system limits. Optimisation is not judgement. More output is not necessarily more value.
The more urgent opportunity may be to use AI to make consequence visible before decisions are made: alongside price, show the energy, material and greenhouse-gas implications of a transaction, while keeping the final judgement human. That would make AI an instrument of stewardship rather than simply a more powerful engine of extraction.
The industrial revolution showed that capability can outrun governance. We should not mistake acceleration for progress a second time.
The author assumes that "..the temptation would never fade: a country that allowed the robots would be vastly richer within a few years. Someone would say yes", and concluded: You can't stop the robot proliferation.
I picked the above "Someone will say Yes" assumption and tried to challenge it. Then I tried to counter-challenge myself. The counter won. This probably means that the assumption is sound.
Challenge: Seeing that such mass production will lead to world instability, there could be international agreements to ban the mass production, import, and transport of human-replacing robots.
Counter-challenge: The world has tried to coordinate on similar problems, such as climate change, over-fishing, narcotics - but it basically hasn't been effective. The combination of profit motive plus geopolitical competition plus social media noisification will likely win.
That's not to say that the author didn't make shaky assumptions. The one about quickly solving the physical intelligence problem is one such.
Thanks Damon,
Your article usefully takes seriously what many economic forecasts avoid: if cognitive work and physical work become automated, production capacity could expand at an extraordinary rate.
MAny (me) would argue the central question is whether an industrial explosion is a sensible objective on a finite planet or indeed a threat multiplier to be avoided
Your article appears to treat the economy largely as a self-reinforcing system of industrial inputs and outputs. Yet the economy is embedded in the Earth system. Energy, materials, land, water, ecological stability and waste-absorption capacity are not merely inputs that can be scaled indefinitely. They are physical conditions, bounded in quantity and constrained by time. Annual doubling of material output is not just a remarkable growth rate; it rapidly becomes an ecological impossibility. Remember the duckweed on a pond that doubles each day?
There is also a question of humans and democracy. If AGI and robotics remove labour as the limiting factor, who determines what is produced, for whom, and to what end? Cheap automated production does not itself distribute shelter, care, security, dignity or political agency. Without deliberate governance, it most likely will concentrate ownership, productive power and the capacity to shape public reality.
Alignment must extend beyond the behaviour of the machine and the few who benifit. A system can be perfectly aligned to an objective and still be catastrophically misaligned with human wellbeing, democratic legitimacy and Earth-system limits. Optimisation is not judgement. More output is not necessarily more value.
The more urgent opportunity may be to use AI to make consequence visible before decisions are made: alongside price, show the energy, material and greenhouse-gas implications of a transaction, while keeping the final judgement human. That would make AI an instrument of stewardship rather than simply a more powerful engine of extraction.
The industrial revolution showed that capability can outrun governance. We should not mistake acceleration for progress a second time.
The author assumes that "..the temptation would never fade: a country that allowed the robots would be vastly richer within a few years. Someone would say yes", and concluded: You can't stop the robot proliferation.
I picked the above "Someone will say Yes" assumption and tried to challenge it. Then I tried to counter-challenge myself. The counter won. This probably means that the assumption is sound.
Challenge: Seeing that such mass production will lead to world instability, there could be international agreements to ban the mass production, import, and transport of human-replacing robots.
Counter-challenge: The world has tried to coordinate on similar problems, such as climate change, over-fishing, narcotics - but it basically hasn't been effective. The combination of profit motive plus geopolitical competition plus social media noisification will likely win.
That's not to say that the author didn't make shaky assumptions. The one about quickly solving the physical intelligence problem is one such.
Thanks for your thoughts, very interesting. I miss a "demand" part here, you seem to assume the demand is infinite.