For businesses, the drivers for AI adoption are increased efficiency and outputs, lower costs, and higher profits. AI is a path to cognitive standardization and scaling. In addition to replicating human work, AI is seen as being capable of doing things people can’t do, such as processing and finding patterns in massive amounts of data. Businesses are in search of a predictable workforce without the complications of costly and finicky humans (labor costs account for a large portion of business expenses and the cost of replacing employees is between one-half and two times the employee’s annual salary), and their labor needs (medical insurance, rest/vacations, promotions, etc.). And, when AI isn’t seen as a replacement for people, AI is seen as a way to lower costs related to employing people, such as AI for hiring (AI video analysis optional), AI for coaching, and AI for managing or supervising employees. Using AI to lower costs is an especially pernicious promise for social institutions that are systematically underfunded, like healthcare and education.

Even in the world of social media, where ad-based economic models generate revenue for both social media companies and some users (influencers like MrBeast and Charli D’Amelio who make millions per year), we see a move toward replacing humans with AI. For example, Meta has created 28 AIs that social media users can interact with (in an extra uncanny turn, Meta “partnered with”–cough, paid, cough–several public figures to use their likeness to “embody” the AIs). These personalities–and other virtual AI influencers that are being created by companies and brands–will likely compete with and negatively impact the income of human influencers (we see a similar phenomenon happening in the adult entertainment industry as well).

There is a certain techno-optimism in the pursuit of efficiency and increased profit by replacing humans with AI. There are few metrics or benchmarks by which we can evaluate how effectively AI replaces human work or makes people more productive. Yet employers (and consumers) seem ready to accept AI automation, even if it means loss of quality and/or other tradeoffs.

What about workers?

It seems inevitable that AI will replace people’s jobs or aspects of their jobs, and we’re told that this replacement will lead to more freedom and more benefits for us (“three day work week!,” “AI will do your mundane tasks like attending boring meetings!” Sure, Idris Elba). But the history of automation of work tells a different story. Incredible innovations in agriculture and manufacturing have not led to improving work conditions or more leisure time for workers. Instead, point out Elway & Reznikoff, those innovations have led to a degradation and devaluing of work conditions and workers’ rights (this is more true in the United States and the UK, and less true in Nordic countries where technological innovation has been linked to improving working conditions and decreasing inequality). Unless measures and/or regulations are put in place to protect workers, we can expect declining compensation, degradation of work conditions, and additional work expectations. Workers are right to be worried about their wellbeing and their livelihoods.

Despite promises that AI will usher in a future where workers are better off, not all jobs will be impacted in the same way, thus increasing the likelihood for growing inequities across industries (a similar “AI divide” is likely across higher-income and lower-income countries). For example, a Pew Research Center analysis placed the following jobs in the category of highest exposure to AI (see also this internationally-focused analysis from the UN’s International Labour Organization):

• Budget analysis
• Data entry jobs
• Tax preparers
• Technical writers
• Web developers

High exposure indicates a strong likelihood of job displacement. As with previous disruptions to jobs caused by technological innovation, workers whose jobs are significantly impacted or rendered obsolete by AI may have limited opportunities to move into jobs that are created by AI. For example, because of the deployment of content moderation AI, which displaces much of the low-wage human moderation on tech platforms, the work of content moderation and reinforcement training will require increasingly specialized skills and will become an increasingly precarious line of work.

All of this raises questions about what agency workers have in deciding what aspects of their work is automated and how AI is used in their work. For example, can a worker refuse to have AI Companion and automated transcription in their Zoom meetings? A version of this question came up during the 148-day Hollywood screenwriters strike in 2023. The strike resulted in a series of AI-focused protections being added to the union contract. It prevents AI from being awarded “first writer” credits, prevents companies from forcing writers to use AI, and prohibits using writers’ material to train AI, in addition to other protections.

Agency requires awareness of what is happening and power to act. As AI technologies become more pervasive in the “technologies of work” –e.g., they’re in our email systems, our finance systems, word processing and document handling, HR systems–they become more difficult to parse and to deal with. Workers can educate themselves about AI but there also need to be commitments from companies to help workers know about and understand how AI could impact their work. Here are some questions to consider:

• Education: What resources or training can be provided to help workers understand AI technologies relevant to their roles?
• Human oversight: Can you promote or create expectations for human oversight of AI implementations? Are there clear channels for questions or concerns about AI-driven processes?
• Transparency & governance: What mechanisms could be put in place to ensure worker voices are heard in the decision making process?
• Accountability: Who is ultimately responsible for the outcomes of decisions made by AI systems in the workplace?
• Ethical considerations: How can organizations ensure that AI technologies align with their stated values and commitment to employees?

Can/will workers have agency and power to shape what happens with AI and work or, like John Cash’s song The Legend of John Henry’s Hammer, will we be mired in a constant struggle against the machine, only to be replaced by the machine in the end?

Views from higher education

Higher education is not immune from the pressures to leverage AI to improve productivity and lower costs. AI is appearing at a time when education, and higher education in particular, is suffering from budget reductions, increased political divergence on its value, and a lower perceived return on investment from college degrees. At the same time, the cost to attend college continues to climb and student loan issues approach a crisis level with 42 million individuals owing a combined $1.7 trillion. These are major societal problems with complex causes. In “Great Ready for Faculty Bot-ification” John Warner focuses on the increasing integration of AI tools in higher education, particularly in teaching roles traditionally held by faculty. Warner feels strongly that AI use is not about faculty productivity but “is instead a direct path to obviating the human role in teaching.”

We asked a couple of colleagues in our organization to reflect on the intersections of AI with their work and how they are approaching decisions about what that looks like for them.

• AI in Instructional Design – Dr. Sarah Payne
• AI in Research and Scholarly Practice – Dr. Sarah Lohnes Watulak