I noticed you mentioned other AI issues in your previous article— misinformation, climate change. It’s likely before we have AI labor protests, we see reforms in reaction to dissent for critical issues that will affect the US more immediately: deepfakes of politicians making inflammatory statements, convincing AI phone call scams, and AI-related environmental impacts.
Do you think it’s possible that general AI dissent could take place that could address issues of AI labor under its umbrella? I see the greatest near-term threat to be convincing deepfakes used for misinformation. Personally, I have family friends who’ve fallen victim to AI misinformation— and this is before anything like the realism of models like Google Veo 3. More immediate issues could facilitate reforms before we hit that critical mass as you discussed.
Also a big factor regarding AI reform: the big beautiful bill. As it currently stands, it ties the hands of states, leaving only the federal government to make the calls regarding AI regulation.
I think deepfakes are so bad that Congress theoretically could do something about them sooner than later, yes, but I don't see mass protests forming about that issue in particular either, or the others aside from job displacement. Of course, I could be wrong!
It's a provocative thought, though, that they could all merge into a general AI backlash protest movement, and that makes some sense. In the current immigration protests, some Waymos have been lit on fire, and there have been other similar incidents before. I think it is possible that these are already signs of a more general backlash, though I haven't looked into that deeply yet. Nevertheless, I think for a larger AI-specific protest movement, you need an anchor, and that anchor is still most likely to be related to job displacement for the reasons listed in the posts.
One approach to dealing with the AI labor crisis is a universal basic income combined with high corporate taxes and high taxes on high income brackets to fund the UBI. If the economy is fully automated, most of America's productive capacity will have consolidated under the control of a few enormous corporations that do not need human labor. As a result, the only people with jobs will be those who control these corporations, their sycophants and a few people who do what cannot be automated. All three groups will be paid extremely well while the rest of America's population will be jobless.
Without interference, America's 99% will get sucked dry and run out of money. The automated corporations will trade among themselves, allowing for economic growth even as the majority of the population becomes destitute.
The function of high, progressive taxes on corporations is to pump wealth out of the automated corporations to be used on a UBI. Similarly high taxes on extremely wealthy individuals would be one of the many measures necessary to prevent the automated corporations from finding a tax shelter they can use to give their highest ranking members more wealth and power. In theory, money would flow from the people to the corporations to the government, which then redistributes the wealth via a UBI back to the people. The new taxes are necessary because a UBI that can fully replace lost jobs would be exceptionally expensive. If America's government implemented a $50,000 basic income available to all adults, it would cost 13 trillion per year.
In practice, it will be very difficult to get corporations and their shareholders to give up their wealth even if the government was controlled by the people. This breaks the loop because the government will not be able to recoup all the money spent on the UBI via taxes, making either inflation via money printing or a debit crisis inevitable.
The society that results from a successful implementation of the UBI I describe has several flaws that I disapprove of. Chief among them is that the average American becomes fully dependent on a government which has drifted away from its founding ideals. Nevertheless, I suspect that the society I describe would be metastable, stagnant and prosperous, serving as a baseline other outcomes can be compared to.
Notably, this is not my preferred solution. My preferred solution revolves around the development of spinal implants that can reroute neural signals to and from external machines on command. Among other benefits, this technology would allow people to directly control machines as if they were their own limbs, reducing the need for AI labor. I would need to make another comment to elaborate and explain further since adding more paragraphs to this comment would make it too unwieldy.
I do think eventually UBI will be needed, but I don't see getting there quickly, that is, in whatever the first wave of policies would be enacted in reaction to initial AI job displacement. I think the long-term result is also interesting and you’ve given me something to think about, so thank you, but in this post I was more thinking what might transpire in the next several years. I suppose I think that I don't see UBI materializing that quickkly because I don't think the larger transformation will unfold that quickly, though FWIW I wrote a bit about that here: https://gabrielweinberg.com/p/is-runaway-ai-coming-in-years-or
In the example above, the Yellow Vest policy enactment was on the order of $150B (U.S.-sized and in USD). As you noted, a reasonably robust UBI is 13T or at least two orders of magnitude higher. Just a $10K UBI for all Americans would be $3T, or about 20x higher. You could make it higher and not for everyone, but we're still talking in the trillions either way.
I think a UBI or some other large, expensive reform will be necessary before the development of AGI. This is because I believe many sectors can be automated with only moderately more advanced versions of today's AI, similar to what Figure is developing.
I believe with significant uncertainty that about 80% of the jobs in the BLS categories of Manufacturing, Wholesale trade, Transportation and warehousing, Leisure and hospitality and Retail trade can be automated. This translates to 46.2 million jobs disappearing. Considering job losses in other sectors, there could be about 60 million jobs in danger of disappearing soon, or about 35% of the workforce. We agree that this shock will be large enough to require a policy at least an order of magnitude larger than the reforms enacted as a result of the Yellow Vest protests. Since America's government will likely balk at the price tag of these reforms, repression becomes likely instead.
One qualifier is that even if AI is good enough that it can replace human labor, there are other factors which may prevent this from happening. For instance, I do not think the robots of today are cheaper than a third world laborer earning dollars per day. As a result, after a few waves of job replacement first world wages could drop down to third world levels, stalling the transition. If the loss of income was slow enough and the propaganda layered thick enough, unrest could be avoided as too many Americans cannot figure out what is going on.
Another set of possibilities is that resource shortages, energy shortages or an American debit crisis destabilize the economy and derail the transition. If this occurs, most of our discussion becomes a moot point. A destabilized economy will shrink so much that even a token UBI becomes unaffordable. Other reforms would similarly become implausible due to cost.
I noticed you mentioned other AI issues in your previous article— misinformation, climate change. It’s likely before we have AI labor protests, we see reforms in reaction to dissent for critical issues that will affect the US more immediately: deepfakes of politicians making inflammatory statements, convincing AI phone call scams, and AI-related environmental impacts.
Do you think it’s possible that general AI dissent could take place that could address issues of AI labor under its umbrella? I see the greatest near-term threat to be convincing deepfakes used for misinformation. Personally, I have family friends who’ve fallen victim to AI misinformation— and this is before anything like the realism of models like Google Veo 3. More immediate issues could facilitate reforms before we hit that critical mass as you discussed.
Also a big factor regarding AI reform: the big beautiful bill. As it currently stands, it ties the hands of states, leaving only the federal government to make the calls regarding AI regulation.
I think deepfakes are so bad that Congress theoretically could do something about them sooner than later, yes, but I don't see mass protests forming about that issue in particular either, or the others aside from job displacement. Of course, I could be wrong!
It's a provocative thought, though, that they could all merge into a general AI backlash protest movement, and that makes some sense. In the current immigration protests, some Waymos have been lit on fire, and there have been other similar incidents before. I think it is possible that these are already signs of a more general backlash, though I haven't looked into that deeply yet. Nevertheless, I think for a larger AI-specific protest movement, you need an anchor, and that anchor is still most likely to be related to job displacement for the reasons listed in the posts.
One approach to dealing with the AI labor crisis is a universal basic income combined with high corporate taxes and high taxes on high income brackets to fund the UBI. If the economy is fully automated, most of America's productive capacity will have consolidated under the control of a few enormous corporations that do not need human labor. As a result, the only people with jobs will be those who control these corporations, their sycophants and a few people who do what cannot be automated. All three groups will be paid extremely well while the rest of America's population will be jobless.
Without interference, America's 99% will get sucked dry and run out of money. The automated corporations will trade among themselves, allowing for economic growth even as the majority of the population becomes destitute.
The function of high, progressive taxes on corporations is to pump wealth out of the automated corporations to be used on a UBI. Similarly high taxes on extremely wealthy individuals would be one of the many measures necessary to prevent the automated corporations from finding a tax shelter they can use to give their highest ranking members more wealth and power. In theory, money would flow from the people to the corporations to the government, which then redistributes the wealth via a UBI back to the people. The new taxes are necessary because a UBI that can fully replace lost jobs would be exceptionally expensive. If America's government implemented a $50,000 basic income available to all adults, it would cost 13 trillion per year.
In practice, it will be very difficult to get corporations and their shareholders to give up their wealth even if the government was controlled by the people. This breaks the loop because the government will not be able to recoup all the money spent on the UBI via taxes, making either inflation via money printing or a debit crisis inevitable.
The society that results from a successful implementation of the UBI I describe has several flaws that I disapprove of. Chief among them is that the average American becomes fully dependent on a government which has drifted away from its founding ideals. Nevertheless, I suspect that the society I describe would be metastable, stagnant and prosperous, serving as a baseline other outcomes can be compared to.
Notably, this is not my preferred solution. My preferred solution revolves around the development of spinal implants that can reroute neural signals to and from external machines on command. Among other benefits, this technology would allow people to directly control machines as if they were their own limbs, reducing the need for AI labor. I would need to make another comment to elaborate and explain further since adding more paragraphs to this comment would make it too unwieldy.
I do think eventually UBI will be needed, but I don't see getting there quickly, that is, in whatever the first wave of policies would be enacted in reaction to initial AI job displacement. I think the long-term result is also interesting and you’ve given me something to think about, so thank you, but in this post I was more thinking what might transpire in the next several years. I suppose I think that I don't see UBI materializing that quickkly because I don't think the larger transformation will unfold that quickly, though FWIW I wrote a bit about that here: https://gabrielweinberg.com/p/is-runaway-ai-coming-in-years-or
In the example above, the Yellow Vest policy enactment was on the order of $150B (U.S.-sized and in USD). As you noted, a reasonably robust UBI is 13T or at least two orders of magnitude higher. Just a $10K UBI for all Americans would be $3T, or about 20x higher. You could make it higher and not for everyone, but we're still talking in the trillions either way.
I think a UBI or some other large, expensive reform will be necessary before the development of AGI. This is because I believe many sectors can be automated with only moderately more advanced versions of today's AI, similar to what Figure is developing.
I believe with significant uncertainty that about 80% of the jobs in the BLS categories of Manufacturing, Wholesale trade, Transportation and warehousing, Leisure and hospitality and Retail trade can be automated. This translates to 46.2 million jobs disappearing. Considering job losses in other sectors, there could be about 60 million jobs in danger of disappearing soon, or about 35% of the workforce. We agree that this shock will be large enough to require a policy at least an order of magnitude larger than the reforms enacted as a result of the Yellow Vest protests. Since America's government will likely balk at the price tag of these reforms, repression becomes likely instead.
https://www.bls.gov/emp/tables/employment-by-major-industry-sector.htm#top
One qualifier is that even if AI is good enough that it can replace human labor, there are other factors which may prevent this from happening. For instance, I do not think the robots of today are cheaper than a third world laborer earning dollars per day. As a result, after a few waves of job replacement first world wages could drop down to third world levels, stalling the transition. If the loss of income was slow enough and the propaganda layered thick enough, unrest could be avoided as too many Americans cannot figure out what is going on.
Another set of possibilities is that resource shortages, energy shortages or an American debit crisis destabilize the economy and derail the transition. If this occurs, most of our discussion becomes a moot point. A destabilized economy will shrink so much that even a token UBI becomes unaffordable. Other reforms would similarly become implausible due to cost.