Par Yossi Sheffi, MIT Professor | Supply Chain, Resilience, & Risk Management Expert
Short-term thinking about AI could have detrimental long-term impacts.
In the past few months, numerous high-profile companies have announced layoffs that they say are related, at least in part, to their adoption of artificial intelligence. Amazon, C.H. Robinson Worldwide, Cisco Systems, Dow, HP, IBM, McKinsey & Co., Nike, UPS, and many more have said that AI is allowing them to reduce headcount or “streamline” how administrative and knowledge work is done.
The volume of layoffs is significant. According to the recruiting and outplacement firm Challenger, Gray & Christmas, internal use of AI was cited as a factor in more than 49,000 announced job cuts in the first four months of 2026, and AI was the leading reason for layoffs at those companies in March and April. And those are just the ones that were publicly announced; undoubtedly there have been many others.
It seems inevitable that there are more to come. After all, AI’s promise of greater efficiency and lower costs is hard to ignore. But decision-makers need to be aware that replacing people with automation on any sort of scale could generate two problems for their businesses: first, it creates dependency on the AI provider(s), and second, it can greatly diminish demand for goods and services.
The Problems of Dependency and Demand
When I say “dependency,” that is not an exaggeration. Companies that shed a large part of their workforce will be at the mercy of their AI vendors, with no alternative to using their products. This opens the way for providers to profit from their customers’ dependency. Venture capital organizations and the AI companies themselves are investing trillions of dollars in a race to develop better and increasingly powerful AI models. At the same time, they are building enormous data centers to power countless transactions. At some point, they will need to recoup these huge sums of money. Corporate users will be charged as much as the market will bear — and if companies come to rely too much on AI, then the market will bear a lot. Changing providers in hopes of reducing costs may not make much of a difference, because, aside from the costs of changing providers, all of them will have to recoup their investments and pay their debts.
This is why companies should not be in a hurry to let their workers (and their expertise and institutional knowledge) walk out the door. Decisions about where and how to apply AI and when human oversight and decision-making add value must be strategic. Short-term thinking without considering the full implications of automation will produce short-term benefits, but if not implemented with care, it could potentially lead to long-term damage in a company’s service quality, customer relations, and reputation. Of course, companies may still deploy AI because it can be faster and more efficient than any human. In addition, once the developers start charging market rates for their AI, human workers may end up being more attractive than the machines.
Some companies are slowing down their hiring, creating a dearth of job openings for many graduates. Again, this should be done in moderation. Hiring and investing in growing human talent ensures the availability of expertise when disaster, such as a cyber-attack, grid overload, or a flood in a data center, impedes digital operations. And even when there are no significant disruptions, these experts will be needed to monitor and keep developing the automation that has been installed.
As for the impact on demand, businesses that successfully deploy AI and consequently lay off employees could end up undermining demand for their own and other companies’ products and services. Yes, the savings will accrue to them, but unemployed people have little or no income to purchase goods and services, and the resulting reduction in demand will affect everyone. Unfortunately, each individual company is locked into the race for efficiency, and leaders may believe that their only option is to run faster, increasing their use of AI and letting more people go.
This is a classic example of the “tragedy of the commons,” where individuals acting in their own self-interest deplete a shared resource, contrary to the common good. The concept was described by Garret Hardin in 1968. Hardin’s example is a town pasture that is open to all. Each shepherd sends more and more animals to the pasture because they can graze there at low or no cost. The problem is that after a while, the pasture becomes bare, the animals do not have enough to eat, and everybody loses.
We are seeing this conflict between short-term self-interest and long-term societal good play out in the pollution of the ocean and the atmosphere, CO2 emissions and global warming, deforestation, the political sphere, and other examples of organizations and individuals prioritizing their own interests above long-term public benefit. I believe it also applies to the AI race, in which replacement of workers by AI financially benefits each company that deploys it but will damage aggregate market demand over time. Progressives argue for government intervention, but unfortunately, most such interventions either did not get off the ground, or they did not yield the hoped-for change in behavior. I hope decision-makers will not allow themselves to be caught up in the AI race without thinking carefully about all potential outcomes – not just now but also in the future.
