AI will not take our jobs (part 2)

Part 1 here:

3. “The lump of labor” fallacy.

The fallacy is well-known enough to have a name. There are actually multiple different fallacies involved here.

3a. Overall demand for goods (and thus labor) rises with automation and spare time/labor.

In a hypothetical competitive market economy, demand is not fixed at “current levels of consumption.” A competitive market economy is not what we have in the US, but bear with me for a second. If a job category of producing widget X suddenly becomes fully automated, sending a lot of people who previously produced X out of work, then the consumers of widget X also suddenly have spare cash to spend on other things. Even if the consumers don’t get the cash because X is built by a monopoly M, then the shareholders of M get the cash. In other words, someone now has the additional spare purchasing capacity they were not using on X and can now use it to buy widgets Y. The people working on X now move on to working on Y. The world has both X and Y now. As mentioned before, this theoretical model has happened in practice since the dawn of time. X used to be food, Y was metal. X was metal, then Y became paper. X was paper, then Y became electricity. I am not being very precise here; I hope you get the idea.

So even if we have individual job categories that disappear from time to time, the spare income and free time of society create the need for more jobs, likely different ones in the future.

3b. We cannot sustain our current standard of living with our current level of work.

In the US right now, there is another reason that “the lump of labor” fallacy doesn’t hold. And that reason is the nearly $40T of debt. This debt is obviously horrifically unsustainable and will have to be paid out in some way that certainly decreases people’s standard of living. Either more taxes (which destroys the incentive to work), a lot fewer benefits (the best solution, but certainly unpopular with some people), money printing (which doesn’t even solve the problem due to the Fed’s structure), or some sort of default or wealth confiscation scheme. None of the options are pretty; they all reduce the debt-fueled standard of living. Automation can “in theory” bridge the gap. And a lot of people, including Marc A, are optimistic. While I acknowledge the “theoretical possibility” of rapid growth, the historical reality is that in the last 25 years, government waste has outpaced living standard gains from automation, and without a fairly historic turn-around, it will continue to do so. So, the reality remains that the most LIKELY outcome is that EVEN IF automation is increasing efficiency, the average person will have to work a lot more for a smaller standard of living, due to the consequences of out-of-control spending and waste. The problem is compounded by the fact that a lot of automation is being directed towards either negative-sum attention sinks (short-term video slop) or the ad economy, which is likely positive-sum, but only barely.

4. We actually want to reduce work.

If the myth of impoverishment following automation is not backed by any theory and it has been proven false time and time again in the last two centuries (if not more), why does it keep coming back?

The first reason is that “jobs” are a “political currency.” Politicians want votes, and they control payment strings. One of the more “legitimate ways” to gain votes is to promise and deliver “jobs” to your voters. These jobs may be real, such as approving a new factory, or they could be “fake jobs,” such as giving money to an NGO whose only task is to ask for more money. “Jobs” are the primary thing that many politicians obsess about (or want the voters to freak out about), and the question of whether these jobs actually produce anything that raises the overall standard of living is completely secondary. Thus, entire discussions have to move towards “jobs” even if they didn’t start out as a primary topic. Whether or not you believe in X-risk, the discussion that goes: “when you say ‘destroy everything,’ you mean take our jobs” is pretty low quality.

The second reason is marketing. AI companies wish to sell you on the idea that LLMs are not just a tool to “augment” the existing worker. Tools to augment workers go for $100 a month, tops. They are tools to REPLACE existing workers, which could compete in price with the workers themselves. Thus, if they are trying to sell the tools to customers, they wish to anchor them on a higher price with “REPLACE, REPLACE, REPLACE” rhetoric, regardless of how true it is.

The third reason is the desire for more government overreach and/or the desire to appear sophisticated and aligned with it when it happens. I’m talking about UBI, of course, a non-functional solution in need of a problem. The problems with UBI warrant their own separate post.

The fourth reason is a hypothetical scenario where there exists a droid with a piece of software that can swap into any simple job, thus rendering any person under a certain IQ threshold “economically irrelevant.” Now, this objection seems to confuse the notion of “economically irrelevant to elites / a specific subsystem” and “economically irrelevant at all.” While a specific portion of the elite economy might in fact ditch all their helpers and servers, the newly unemployed can still trade with each other in a closed-loop economy. It’s the same reason that Cuba can actually exist. Even if the US moves to extreme automation, but there is zero trade or travel between the US and Cuba, this wouldn’t impact the Cuban economy a whole lot. Same with the low-IQ populations, who can form an “internal Cuba” and trade with each other. The Amish are an example. The more isolated the group is, the easier it is. It sounds “unequal,” because it is, but as long as the two societies don’t mix too much, this situation is relatively peaceful and economically stable. “Trading within your low-tech community” is “economically irrelevant to the elites,” but is completely not the same as being “economically irrelevant at all” and dying on the street.

And the last reason is bad philosophy of what is good / what we actually want. Individuals are anchored to the system via their specific jobs and how much one gets out of them. However, the overall standard of living of society is anchored to how much these jobs produce in surplus value (how much you put in - how much you get out).

This simply means we want to work as little as possible and produce as much as possible, assuming there is stability from an environmental/political/renewable perspective. This is good. We don’t have enough people simply declaring “this is good.” Maybe it sounds “lazy” or “unrealistic,” even though this has been happening for thousands of years. Maybe there is a deep-seated fear of saying “stuff is good,” similar to a deep-seated fear of writing “non-dystopian science fiction.”

The question of distributing this surplus value matters. We cannot simply subsidize older people for infinity and leave young people with no free time for family. We also cannot make up nonsensical income schemes such as UBI. At the end of the day, we WANT to work less for a HIGHER standard of living. Or *at least* for a standard of living which enables the fertility rate to rise above 2 again. After which, we can begin to think about more ambitious megaprojects.