Artificial intelligence is everywhere. But behind the promise of autonomy lies the more complex reality of AI’s dependence on humans. My recent research explores this often overlooked fact and shows that people remain central to how AI works, far beyond being portrayed as passive users and as active contributors.
In a recent MIS Quarterly (open access) publication, I led research on algorithmic autonomy in ride-hailing platforms like Uber. These systems are often powered by algorithms that dispatch drivers, set prices, and optimize routes in a seemingly seamless manner. But beneath the surface, it turns out, there are more subtle differences. This is not a single algorithm making decisions, but a network of algorithms that delegate activities to humans (drivers, passengers, platform workers) in a distributed and sometimes almost invisible way.
This phenomenon is called distributed delegation. Rather than replacing human performance, the platform embeds and redistributes it into interface design, evaluation, and optimization rules. The result is a hybrid behavior where algorithms and humans are codependent. While machines rely on human data, choices, and effort to function effectively, they also shape how humans act and what they can do. These findings challenge the dominant narrative about AI autonomy. We make it clear that algorithms do not automate human involvement. Rather, they reconstruct it, often in subtle ways.
Based on these insights, I am currently conducting research investigating humans and algorithms working together in multifaceted interactions. Rather than asking who is in charge, I focus on revealing how the capacity to act emerges through interactions between actors. I am researching the case for an artificial pancreas system that combines continuous glucose monitoring, an insulin pump, and user-configured algorithms to automate insulin administration. This case provides clear insight into the idea that actions in AI-enabled systems are not performed individually by actors, but emerge from continuous interactions between humans, bodies, algorithms, data, devices, and the environment.
AI promises speed, scale, and accuracy across all industries. However, my research shows that its success still depends on the role of humans. In both platform work and medical settings, humans interpret, modify, and adapt algorithmic systems in ways that shape outcomes.
Designing jobs for humans and AI
For business leaders, this means they need to move beyond the myth of full automation. The challenge is not just how to deploy AI, but how to design and manage human-AI interactions and support people in taking the actions and decisions that make intelligent systems work. As algorithms take on more decision-making functions, workers are often tasked with invisible, fragmented, or reactive forms of labor, such as interpreting machine output, fixing edge cases, and providing flexibility that automated systems cannot provide. This labor is essential, but it often goes unrecognized and unsupported. My research shows that assessing and designing for this hidden work is a key leadership challenge. We need to rethink how we train, support, and reward workers who now work with machines.
For people just entering the workforce, my research carries a hopeful but strategic message. That means the AI still needs you. The most effective systems are not completely autonomous, but rely on human insight, adaptability, and judgment. As intelligent technology becomes more integrated into business processes, your value lies in your ability to operate and work around machines, interpreting output, asking difficult questions, and making contextual decisions. Keeping up with AI shouldn’t be your only goal. Instead, develop human skills that machines cannot imitate: critical thinking, relational sensitivity, and systems-level awareness. These are features that will increasingly be sought after in the workplace of the future.
As AI systems become more prevalent, the real challenge is not replacing human intelligence, but understanding how to build with it. My research in this area continues to highlight how digital systems and human contributions are deeply intertwined. For graduates and future leaders, there is an opportunity to shape technology and organizations that not only automate but emphasize human strengths. A more intelligent future isn’t just about better-thinking machines, it’s about systems that help us all think, work, and live smarter.
About Dr. Marta Stelmaszak Rosa
Marta Stelmaszak Rosa is an assistant professor of operations and information management in the Isenberg School of Management at Amherst, Massachusetts. Her research focuses on digital data and its responsible, sustainable, and socially just management in organizations. She teaches courses on business intelligence and data analysis. Stelmaszak Rosa’s work on designing human-AI interactions shapes Isenberg’s evolving curriculum, from the newly launched undergraduate and graduate AI certificates to expanding elective options across degree programs.
© Copyright 2025 Poets & Quants. All rights reserved. This article may not be republished, rewritten, or otherwise distributed without written permission. To reprint or license this article or any content from Poets & Quants, please submit your request here.
