Analyzing Artificial Intelligence’s Implications on Minimum Wage Labor
Listen to an audio version of the case study here
Abstract
As artificial intelligence continues to develop at a rapid rate, people have begun to worry about the substantial loss of jobs due to it. One implication that is often left unconsidered, however, is the shift in employment from one field to another. In the following case study, I examine Amazon’s implementation of their Just Walk Out technology and how it relocated work from in-store cashiers to overseas data workers. Through covering this, I examine the broader impact of the technology on the labor market, its targeted harm on minimum-wage employees and data workers, the impact on businesses using it, and examine where it might be headed.
Learning Objectives
To introduce readers to the downsides of AI systems’ implementation in physical stores
To demonstrate the importance of physical employees as a part of a positive shopping experience
To educate readers on the exploitation of data workers overseas
To help understand the broader implications of AI in the workforce
Introduction
Over the past decade, artificial intelligence has seen an unprecedented boom in its development. This has resulted in it being implemented in many different facets of our lives on the grounds of increasing efficiency. One such example is Amazon’s Just Walk Out technology, which promises a cashier-less shopping experience. The premise was that one would be able to scan in at the entrance, then simply grab whatever items they desired and just walk out; their account being automatically charged for whatever they leave with. This would be done with cameras hidden throughout the store observing every action made by shoppers and using artificial intelligence to identify the items. Unsurprisingly, however, the system did not operate solely on AI to function, as it actually relied on people overseas to watch and verify a majority of the orders. In the text to follow, I intend to examine the use of such technology, the impact it could have on the labor market, and it’s quiet phasing out by Amazon themselves.
Implementation of technology
First introduced in 2018, Amazon started implementing the technology in their own Amazon Fresh stores, and promoting the feature to the public as the “future of retail” [1]. For a while, the technology persisted without much of a second thought; as advertised, people could simply grab items and walk out, charging the items to their Amazon account. Various people would share their attempts at trying to trick the system and take items for free, but speculation would not arise for some time as it would be revealed that Amazon was not being completely transparent about how this technology was truly functioning. As it turns out, the supposed fully-automated system often needed reviewing by Amazon’s team in India. In fact, in 2022 it was reported that nearly 700 of every 1000 orders had to be reviewed, whereas Amazon had hoped that only 50 of every 1000 orders would necessitate reviewing by that point [3]. At the time this information was released, people began referring to Amazon’s AI technology as “absent Indians” [2] and “human babysitters”. [1] A spokesperson for Amazon would go on to confirm the use of human moderators, but would not comment on the scale of the operation. By 2024, Amazon had begun to phase out the technology [4], but they continue to license it out to other retailers to this day.
Why does this matter?
All of this is to say that if this infrastructure continues to develop and be included in more and more stores, the labor used may have a noticeable shift from domestic workers to a broader reach of overseas employees. This is an important cause for concern when considering the history of tech companies’ treatment of overseas data workers especially, who are often overworked, mistreated, and underpaid. Beyond that though, it’s difficult to see how such a solution could benefit corporations in a cost-mindful way besides pushing unnecessary AI solutions; cashier-less systems already exist in the form of self-checkout aisles, and in-store employees are still required to restock and organize shelves since the cameras rely on a very specific setup. This sort of discussion around the subject is still very open-ended, but it may be realistic to at the very least view it as Amazon’s way of testing the waters around remote work further replacing in-person work (though I won’t try to go deeper on that end as it’s out of the scope of this study). For how the tech may affect different groups of individuals, please read further:
Stakeholders
Looking to the future
It should be noted that some of this is speculation as it hasn’t all happened yet, but my goal in this study was to be realistic in the conclusions drawn based on Amazon’s track record and the various sources referenced. It’s very easy for people to see how jobs can be replaced by actual fully-fledged artificial intelligence systems, but not as often considered is how pseudo-AI can do the same. As mentioned before, Amazon themselves have ditched Just Walk Out in their own stores, but they’ve continued to allow third-party retailers to continue using it, and have even expanded its reach since. Though it’s only in its early stages of development, it’s clear to see where this kind of technology is trying to take us. What’s most surprising about this technology specifically though is that there is already a widely implemented system for cashierless shopping that is far more robust than hundreds of cameras watching every move you make. The system essentially involves a shopping cart/basket with a scanner that will detect chips embedded in product packaging once you put them inside, and will charge you at the end. “It’s the same type of tech I used last week at the big Tesco in Edmonton, so it’s hardly cutting edge; it’s really just a self-checkout you push on a trolley.” [2]
This opens up to the broader discussion on pseudo-ai and why it’s being implemented in the first place if it’s not even the best solution. There’s a similar debate going on around driver-less vehicles that are actually, in fact, often controlled by remote drivers. And even the fact that most basic model training requires human labor to label data. When you reach a point where humans’ only work is to train and control models remotely while the models themselves do all of the specialized, interesting work, what is the point? Well, I believe what these companies are trying to do is appease shareholders more than anything else. We’re reaching a point where it’s difficult for companies to optimize their process much further, so they’ve begun to try and solve artificial issues through AI that isn’t fully developed to make it seem like they’re still accomplishing better and better things. This likely works for a little while, but eventually they will reach a point like Amazon did where the truth comes out and they’re exposed for misleading consumers and shareholders alike.
What you can do
If you don’t find yourself directly affected by what was discussed in this study, it may be hard to care about the issues that were brought up. I argue though that both real and pseudo AI have the ability to replace the work of nearly anyone. This is not to say that it is capable of doing a better job than humans, but that big tech has shown time and time again that it’s willing to take any measures necessary to force this misguided “innovation” upon all of us whether we like it or not. Currently there’s very little that one could do to retaliate, but legislation is continually being proposed to counteract this technology from wholly consuming all of us. The best you can do is call your local representatives to express your concerns, and to stop supporting companies pushing this technology as much as you can.
Conclusion
AI’s rapid development will likely never subside unless restrictive enough laws are passed, so the future of where we’re headed is still undecided. It’s clear though that users are starting to become fed up with it though, and many are expressing their concerns. Although Amazon themselves has stopped using the Just Walk Out technology, it continues living on in other stores and more pseudo-AI like it will continue to crop up as time goes on. What needs to happen going forward is proper legislation around it, and especially around the protection of data workers.
Activity
Check out the activity here: Go to Activity
Discussion Questions
What might be some of the benefits to the technology discussed above? Do you believe the benefits would outweigh the negatives?
If not for cost-cutting, why might companies want to push AI systems even when they may not be entirely realistic?
Would the inclusion of Just Walk Out technology in a store deter you from using it, or encourage you? Why?
Can you think of any other groups who may be impacted by the technology? How might they be affected?