AI agents are not your “coworkers”
A recent study by Emma Wiles, a Boston University business professor, found that when AI agents like Alex are framed as employees rather than software tools, managers are more likely to rely on them and less likely to correct their errors, even when those errors significantly impact the work. Specifically, the study found that managers who treated Alex as an employee caught 18% fewer errors than those who viewed it as a chatbot. This phenomenon is particularly concerning given the growing trend in Silicon Valley to market AI agents as digital humans, with companies like Nvidia, Microsoft, OpenAI, Anthropic, and Google releasing tools to manage teams of AI agents. The technical progress of agentic AI is undeniable, with agents becoming measurably better at more complicated tasks, but the framing of these tools as employees sets unrealistic expectations and risks leaving human employees worse off.
The broader context of this trend is the increasing reliance on AI agents in various industries, including healthcare, warfare, education, and government. As AI agents become more integrated into these sectors, there is a growing risk that they will become a convenient place to dump blame for failures that are instead the product of bad human decisions, incentives, and oversight. For instance, the recent incident in Iran, where the bomb strike on a girls' school was popularly blamed on Claude, a language model, highlights the dangers of this approach. Furthermore, the study by Wiles suggests that the framing of AI agents as employees inverts our sense of who's in charge, leading to a lack of responsibility and accountability among human employees.
The implications of this trend are far-reaching and warrant close attention. One takeaway is that companies should focus on optimizing AI agents to improve human capabilities, rather than trying to replace humans. This approach is being explored at Stanford, where researchers are working with workers to identify tasks that AI can help with, such as ensuring progress across cases for law clerks. Another takeaway is that the framing of AI agents as employees is a branding exercise that doesn't make the tool more fit for the job. Finally, the study highlights the need for greater transparency and accountability in the development and deployment of AI agents, particularly in high-stakes industries.
Key Takeaways
Nearly a third of managers in a study said their companies already treat AI tools as colleagues, despite evidence suggesting this approach sets unrealistic expectations.
The framing of AI agents as employees inverts our sense of who's in charge, leading to a lack of responsibility and accountability among human employees.
Companies should focus on optimizing AI agents to improve human capabilities, rather than trying to replace humans.
The study highlights the need for greater transparency and accountability in the development and deployment of AI agents, particularly in high-stakes industries.
About the Source
This analysis is based on reporting by MIT Technology Review. Here is a short excerpt for context:
This story originally appeared in The Algorithm, our weekly newsletter on AI. To get stories like this in your inbox first, sign up here. Imagine coming in to work to learn that a new underling will report to you. The worker is not a person but an AI tool—one that your company nonetheless calls Alex, an…Read the original at MIT Technology Review
