While the state of discourse may have people feeling like AI is an inevitable fixture of our future, ultimately, people are making the decisions around how and where that tech is deployed. The role of human decision-making, and the sociological and psychological factors behind it, is perhaps understated in conversations around technology adoption today, even as change management shows up as a common barrier to its progress.
The human behavior that guides corporate decisions and working conditions is the research focus of Hatim Rahman, professor of management and organizations and sociology at Northwestern University. His latest book, Inside The Invisible Cage: How Algorithms Control Workers, highlights the role of digital platforms in the modern employee experience and won the 2025 George R. Terry Book Award presented by the Academy of Management.
"In 2021, over forty million people used an online labor platform to find work in the United States. To put that number in context…if the five largest occupations in the United States were combined, that still would not equal as many people as those who use online labor platforms to find work,” Rahman wrote in a published excerpt. “These platforms have transformed how organizations and workers find and work with each other. The goal of these platforms is to create…instant access to top talent from around the world with the click of a button.”
In an interview with Newsweek, Rahman pointed to multiple examples across industries that underscore the importance of professional groups, labor unions and other structures of worker-led power to address the concerns around responsible adoption of emerging technology. For example, writers and creators are fearful of losing their intellectual property; lawyers and health care professionals are protective about the standards of their services.
Those working in retail and manufacturing jobs are more susceptible to management whims unless they have strong representation. Depending on the power structures of those industries, AI has been used in different ways, Rahman found.
“There are normative beliefs that inform the way that people design [new tech systems] and that carries over to the testing and then the implementation, so by the time the worker gets it, there's been all these things that are already baked into it that heavily influence it,” he said.
Some examples are more high-profile than others, but the strength of professional advocacy groups plays a powerful role in the state of AI adoption in industries like health care, education, legal and other professional services firms and the entertainment industry.
“There were ideologies, incentives and interests that influence the way that [AI technology] was developed and deployed,” Rahman explained. “Shining a light on that has a huge impact on thinking about the effects of inequality and where to potentially mitigate them.”
Rahman’s recent paper studied the impact of AI on workplace inequality, identifying a cross-disciplinary view of the subject based on economic factors like wages or career progress, but also developmental and structural inequality.
“Economic literature is interested in wage inequality, whereas the social and psychological [researchers are] interested in cognitive inequality,” Rahman said.
His research also calls into question the accepted views on productivity gains in the AI ROI conversation, in that having a bot complete a task may also hinder employees’ development.
“Sometimes, if you just look at the task [being] completed more efficiently…you may see increases. But in the long term, if workers are not gaining the requisite skills through it as well, then it's unclear how sustainable those productivity gains are,” he explained.
Using AI as a task or job replacement, or measuring the impact of technology with just a meter on time saved, still faces a major hurdle: what happens in the event of a system shutdown or failure? There is a cost to dependence on technology.
“If the system goes down, which we're increasingly seeing happening, what does that mean for the economy? Does the organization shut down?” Rahman asked. Employees may also be less equipped to fix or workaround system errors if they’re overly reliant on automation.
If generative AI or automation is deployed as a cost-saving measure in marketing, customer service or the company’s products, without proper guidelines for its use, it could lead to service failures or decreased sales or customer loyalty. In hiring and performance management, it can propagate bias, or as Rahman's book details, exert control over working conditions in what he calls, "an enhanced form of Taylorism," with opaque standards hidden behind the validity of hot, new tech.
"As organizations increasingly use algorithms to make consequential decisions for people inside the invisible cage (such as deciding who can rent and buy property, who goes to jail, and whom to hire)," he wrote in the book excerpt. "This form of control increasingly determines our opportunities without allowing us to understand or respond to the factors that govern our success."
Request Reprint & LicensingSubmit CorrectionView Editorial & AI Guidelines
Add Newsweek as a preferred source on Google to see more of our trusted coverage when you search.Recommended For You
Newsweek.AIAI Impact: The Limits of Scale10 min read
Newsweek.AIWorkera is Enabling Skills-Based Talent Strategies Using AI5 min read
NewsThe Day AI Got My Clients7 min read
Newsweek.AICarlos Sainz’s F1 Car Wears AI-Inspired Livery at Las Vegas GP2 min read
Newsweek.AIA Market Correction, Not a Meltdown, Is Hitting AI5 min read
Newsweek.AIThe Bittersweet, Forrest Gump-like World of AI, According to Gary Marcus22 min readTrending
National Weather ServiceWinter Storm Alert Issued as 12 Inches of Snow Set To Hit3 min read
Higher EducationFull List of Degrees Not Classed As ‘Professional’ by Trump Admin8 min read
MoneyNew ‘Cash Law’ Could Change Grocery Store Rules3 min read
LawsuitCampbell’s Soup VP Mocks ‘Poor People’ Who Buy Its Food in Secret Recording6 min read
ThanksgivingHome Depot Responds to Thanksgiving Boycott5 min read
