This week we talk about robots, call center workers, and convenience stores.
We also discuss investors, chatbots, and job markets.
Recommended Book: The Fourth Consort by Edward Ashton
Transcript
Though LLM-based generative AI software, like ChatGPT, Gemini, and Claude, are becoming more and more powerful by the month, and offering newfangled functionality seemingly every day, it’s still anything but certain these tools, and the chatbots they power, will take gobs of jobs from human beings.
The tale that’s being told by upper-management at a lot of companies makes it seem like this is inevitable, though there would seem to be market incentives for them to both talk and act like this is the case.
Companies that make new, splashy investments in AI tech, or which make deals with big AI companies, purporting to further empower their offerings and to “rightsize” their staff as a consequence, tend to see small to moderate bumps in their stock price, and that’s good for the execs and other management in those companies, many of whom own a lot of stock, or have performance incentives related to the price of their stock built into their larger pay package.
But often, not always, but quite a lot of the time, the increased effectiveness and efficiencies claimed by these higher-ups after they go on a firing spree and introduce new AI tools, seem to be at least partly, and in some cases mostly attributable to basically just threatening their staff with being fired in a difficult labor market.
When Google executives lay off 5 or 10% of their staff on a given team, for instance, and then gently urge those who survived the cull to come to the office more frequently rather than working from home, and tell them that 60 hours a week is the sweet spot for achieving their productivity goals, that will tend to lead to greater outputs—at least for a while. Same as any other industry where blood has been drawn and a threat is made if people don’t live up to a casually stated standard presented by the person drawing that blood.
Also worth mentioning here is that many of the people introducing these tools, both into their own companies and into the market as a whole, seem to think most jobs can be done by AI systems, but not theirs. Many executives have outright said that future businesses will have a small number of people managing a bunch of AI bots, and at least a few investors have said that they believe most jobs can be automated, but investing is too specialized and sophisticated, and will likely remain the domain of clever human beings like themselves.
All of which gestures at what we’re seeing in labor markets around the globe right now, where demands for new hires are becoming more intense and a whole lot of low-level jobs in particular are disappearing entirely—though in most cases this is not because of AI, or not just, but instead because of automation more broadly; something that AI is contributing to, but something that is also a lot bigger than AI.
And that’s what I’d like to talk about today. The rapid-speed deployment, in some industries and countries, at least, of automated systems, of robots, basically, and how this is likely to impact the already ailing labor markets in the places that are seeing the spearpoint of this deployment.
—
Chatbots are AI tools that are capable of taking input from users and responding with often quite human-sounding text, and increasingly, audio as well.
These bots are the bane of some customers who are looking to speak to a human about some unique need or problem, but who are instead forced to run a gauntlet of AI-powered bots. The interaction often happens in the same little chat window through which they’ll eventually, if they say the right magic words, reach a human being capable of actually helping them. And like so many of the AI innovations that have been broadly deployed at this point, this is a solution that’s generally hated by customers, but lauded by the folks who run these companies, because it saves them a lot of money if they can hire fewer human beings to handle support tickets, even if those savings are the result of most people giving up before successfully navigating the AI maze and reaching a human customer support worker.
In India right now, the thriving call center industry is seeing early signs of disruption from the same. IT training centers, in particular, are experimenting with using audio-capable AI chatbots instead of human employees, in part because demand is so high, but also, increasingly, because doing so is cheaper than hiring actual human beings to do the same work.
One such company, LimeChat, recently said that it plans to cut its employee base by 80% in the near-future, and if that experiment is successful, this could ripple through India’s $283 billion IT sector, which accounts for 7.5% of India’s GDP. Hiring growth in this sector already collapsed in 2024 and 2025, and again, while this shift seems to be pretty good for the balance books of the companies doing less hiring and more firing as they deploy more AI systems, it’s very not good for the often younger people who take these jobs, specializing in call center IT work, only to find that the market no longer demands their skill sets.
Along the same lines, but in a perhaps more surprising industry, some convenience stores in Japan are deploying robots to manage their back rooms, where the products that end up available out front are unloaded, tallied, and shelved.
These robots, which are basically just arms on poles, sometimes attached to wheeled bases, for moving around, sometimes not, are operated by AI, but are also continuously monitored by human employees in the Philippines. Each worker, who can be paid a lot less than an entry-level, young Japanese person would expect to be paid, monitors about 50 machines at a time, and steps in, using virtual reality gear to control the robots, if one of them gets stuck or drops something; which apparently happens about 4% of the time.
This is akin to offshoring of the kind we’ve seen since the early 2000s, when the dawn of technological globalization made China the factory of the world and everything shifted from a model of local production and the stockpiling of components, to a last-minute, supply-chain oriented model that allowed companies to move all their manufacturing and some of their services to wherever it could be done the cheapest.
Many people and companies benefitted from this arbitrage to some degree, though many regions have dried up as a result of this shift, because, for instance, former company towns where cars were produced no longer have the resources to keep infrastructure from degrading, and no longer have enough jobs to keep young people from moving away; brain drain can become pretty intense when there’s no economic reason to stay.
This reality is expected to become more widespread, even beyond former manufacturing hubs, because of the deployment of both AI systems, which can be subbed-in for many remote jobs, like call center work, programming, and the like, but also because of increasingly sophisticated and capable robots, which can do more automated work, which in turn allows them to be monitored, sometimes remotely, like those Japanese convenience store robots, for a fraction of the price of hiring a human being.
This shift is expected to be especially harrowing for teens hoping to enter the labor market in entry-level jobs, as responsibilities like shelf-stocking and product scanning and the loading and unloading of materials are increasingly automatable, as robots capable of doing this work are developed and deployed, and perhaps even more importantly, as systems that augment that automatability are developed and deployed.
In practice, that means coming up with shipping processes and other non-tangible systems that lean into the strengths of today’s automated systems, while reducing the impact of their weaknesses.
Amazon is in prime position to do exactly this, as they’ve already done so much to rewire global shipping channels so that they can deliver products as rapidly as possible, to as many places as possible. As a result, they control many of the variables within these channels, which in turn means they can tweak them further, so that they’re optimized to work with Amazon’s specialized automated systems, rather than just human ones.
The company has stated, in internal documents, that it plans to automate 75% of its total operations, and it currently has nearly 1.2 million employees. That’s triple what it employed in 2018, and it’s expected that the automated systems it has already and will soon deploy will allow it to hire 160,000 fewer people than planned by 2027.
Even though the company expects to sell twice as many products by 2033, then, it expects to hire 600,000 fewer people by that same year. And it’s so confident in its ability to make this happen that it’s already making plans to rebuild its image in the aftermath of what’s expected to be a really difficult period of people hating it. It’s planning significant branding efforts, meant to help it seem like a good corporate citizens, including sponsored community events and big donations to children’s programs.
It’s also intending to frame this shift as an evolution in which robots are amplifying the efforts of human employees. Rather than calling their automated systems robots, they might call them ‘cobots,’ for instance.
Amazon has contended that the internal documents in which these plans were outlined, those documents acquired and reported upon by the New York Times, are incomplete and not an accurate representation of what Amazon plans, and they said those branding efforts are not a response to hate related to their automation efforts, they just like spending money on nice things for communities.
The net-impact of existing efforts of this kind, though, is to deplete local job markets where these big companies dominate, and to make the jobs that survive a lot higher-end, requiring more technical sophistication, often, like being able to manage and maintain these sorts of robots, which are skills few people currently have.
Amazon’s backend is already very automated, powered by bots originally developed by robotics maker Kiva, which was purchased by the company for 3/4 of a billion dollars back in 2012. Amazon warehouse workers now work alongside all sorts of robots—though as seems to be the case with employees who survive AI-related firings, those humans who remain are often subjected to strenuous conditions and a lot of pressure to work long hours.
In the company’s Shreveport, Louisiana location, there are more than a thousand robots working around the clock, and that’s allowed Amazon to hire 25% fewer human workers at that facility, while processing 10% more items. The plan is to further refine that model while also spreading it to other Amazon warehouse locations, 40 more of them by 2027, which is part of how they expect to reach that aforementioned 75% employee reduction goal.
Amazon’s obviously at the forefront of this shift because of the nature of their business and business model, but other big employers, such as Walmart, are also pushing in this direction. Walmart officials have said they will have cut costs by more than 30% at facilities where they’ve been experimenting with more automation by the end of 2025, and they’ve already cut those costs by 20% at these facilities, in part because fewer human employees are necessary.
All of which is interesting in part because these are clearly real innovations that are leading to more efficiency and effectiveness at lower costs, and ultimately these may translate into cheaper goods and services for customers if the companies deploying automated technologies decide to pass on those savings.
But simultaneously, this represents a fundamental shift in the job market and overall economy, and if new jobs don’t arrive at the same scale and pace as they’re disappearing, or some other money-distribution solution, like a minimum basic income, doesn’t arrive in time, we could find ourselves in a situation, globally, but especially and most immediately in markets like China, which has far more automation than everyone else right now, and the US—we could have a situation where there’s just a whole lot of stuff being made, but not enough people who can afford it, because they can’t find jobs that will pay them enough to participate in the economy, which in turn could splashback on these automated measures in a negative way, as these companies’ addressable markets shrink.
Show Notes
https://collectivefutures.blog/the-infrastructure-of-meaninglessness/
https://www.cnbc.com/2025/02/20/meta-approves-plan-for-bigger-executives-bonuses-following-5percent-layoffs.html
https://www.nytimes.com/2025/02/27/technology/google-sergey-brin-return-to-office.html
https://www.reuters.com/world/india/meet-ai-chatbots-replacing-indias-call-center-workers-2025-10-15/
https://restofworld.org/2025/philippines-offshoring-automation-tech-jobs/
https://www.theverge.com/report/806728/tech-left-teens-fighting-over-scraps-robots-taking-jobs
https://www.theverge.com/transportation/805471/waymo-robotaxi-winter-snow-weather-testing
https://www.wired.com/story/elon-musk-wants-strong-influence-over-the-robot-army-hes-building/
https://www.supplychaindive.com/news/walmart-automation-supply-chain-cost-savings/747377/
https://www.nytimes.com/2025/04/23/business/china-tariffs-robots-automation.html
This is a public episode. If you'd like to discuss this with other subscribers or get access to bonus episodes, visit letsknowthings.substack.com/subscribe