A storm is coming. While you are being distracted with AI-generated memes and porn, while laughing at examples of ChatGPT”s inability to reason properly, AI and automation in general are already in the process of replacing countless workers. I am not an AI cheerleader. In fact, I consider extensive and uncritical AI use a marker for low intelligence. I cringe when I see an email or document that was obviously written by an AI, and the human sending it out was unable to notice and fix the glaring shortcomings of the text. You are not going to boost your IQ with AI. However, if you are smart you will be able to multiply your output, depending on the task, of course.

An interesting observation I recently made was that tech giants like Amazon and Microsoft seem to have hit peak human workers. Amazon, for instance, is shedding not just office drones but also warehouse workers. The number of parcels processed per human warehouse worker is skyrocketing, because robots and intelligent routing algorithms have become so much more effective. The number of robots they use is also roughly the same as the number of humans. More and more warehouses, in general, are fully automated. There is even a term for it: “dark warehouse”. They are dark because the absence of humans means that there is no need for artificial lighting. Robots just keep working around the clock. Of course, there is the bullshit hand-wringing of executives, like the Amazon guy in the WSJ article linked to above, who says that, “the company will continue to need many workers and that new robots are meant to make their jobs easier, not displace them.” Call me a cynic, but I cannot quite imagine a future in which Amazon uses countless robots just so that their existing workforce can lean back, sip Mountain Dew and watch the robots work. A lot of people are going to lose their jobs, and a lot of people already have lost their job due to AI and automation. Interestingly, there was a time, a mere three years ago, when Amazon was afraid they would use up all the available local low-skilled workforce. They no longer have this concern.

There are few jobs where low-skilled workers are needed in large numbers. Working in a warehouse is an example. Personal transportation is another. On that note, I recently read about Waymo scaling up their fleet. They currently do 250K fully automated rides per week. Using a simple back-of-the-envelope calculation, this tells me that they have replaced already over 1,000 full-time drivers, i.e. 250,000 trips * 10 minutes / 60 (min/hours) to get the total hours, and divided by 40 hours (per week) to get the number of drivers, assuming the highly unrealistic assumption that there is no downtime. In reality, Waymo has probably replaced at least 5,000 cab drivers already. If you think that these are not a lot of people, keep in mind that Waymo is currently only in four cities in the US. Imagine they roll out their services to every city greater than 100,000 people! I do not see why you could not just replace almost every cab driver this way, with the small exception of high-end chauffeur services. What will these people do instead?

In the most recent All-In podcast there was a segment by Travis Kalanick on fully automated commercial takeaway kitchens. Obviously, these do not serve the rich who will keep frequenting Michelin-starred restaurants, but such technology will have a massive impact on the labor market as this kind of work was another option for the low-skilled sector. According to Travis Kalanick, the cost of such a setup is 1/4 of a regularly staffed kitchen. This is game over for these jobs.

Going further up the skill hierarchy, I can share some insights from the world of software development. You may have learned that the market is no longer as hot as it used to be. These days, a degree in computer science makes a fresh college graduate less employable than a degree in History. This will only get worse. Entire categories of tasks can nowadays be done more effectively by a good software engineer who uses AI to do the bulk of the work. I am not talking about some nonsensical “vibe coding” but large-scale migration tasks, i.e. rewriting your code in a different programing language or a more recent version of the same programming language. This may take months if done manually but with AI it is a matter of a few weeks. From guys in testing I hear that work that would normally take two to three weeks can be done in half a day. Even regular development is affected, particularly the boilerplate-heavy frontend area.

The consequence of the advancements in AI in software engineering is that the run-of-the-mill developer will have a really hard time justifying his or her existence. Important to point out is that very little commercial work is so-called greenfield. Normally, people maintain existing software, which means making a few adjustments on the frontend, doing a bit of backend work, making sure the test suite is up-to-date, and keeping their code current. Very often, this i the bulk of the work and AI can take care of this quite well already.

In the West, we are currently seeing first signs, at the level of the elites, of a pushback against unlimited mass-immigration, both low and high-skilled. I believe that this is a reaction to market changes. However, this is too little, too late. I see AI usage getting normalized, but not in an idiot way, meaning that people believe they can outsource their entire day job to ChatGPT, but instead AI tools are used as multipliers. A mediocre developers with an AI-coding assistant will only write more garbage code than ever before but a skilled developer can create a project that would otherwise take an entire team basically by himself. This is currently happening and more and more companies are jumping on this bandwagon. In industries where hardware or equipment is the bottleneck, the adoption may be slower, but it is nonetheless happening. It is happening way faster than the elites may realize.

The dystopian version of the aforementioned scenario is that we will have many millions of people who used to work in warehouses or made a living as cab drivers being put out of work, unable to find any other legitimate way of making money. We have already automated away a lot of low-skilled work, so what are they going to go? To preemptively address the Reddit galaxy brain take that we have always found ways to use the available workforce: This has not been the case for over half a century! In the West we have experienced persistent unemployment since at least the 1970s. Governments simply got creative with hiding the true magnitude, for instance by shoving everybody into a four-year degree program or even downright statistical manipulation. A while ago I read that Denmark readily declares people who are unable to find a job as “handicapped”, in order to keep unemployment numbers down.

According to a quick AI search, there are around 2 million Uber drivers in the US, and over 15 million restaurant workers. The number of software engineers on an H-1B visa is supposedly only around 700,000, but this does not seem plausible, based on what I read about the US software market, where apparently well over half of all employees, in an industry of over 4 million employees, do not have WASP credentials. These are just three industries. Give it a few more years and there will be a lot of questions about what to do with all these people who lost their job due to AI and automation. This could be one angle for a strong right-wing push for mass deportations, albeit I am not quite sure how this should happen in the US. Without it, though, I think there will be enormous social upheaval.