I just read a new study from Stanford called “The Future of Work with AI Agents,” and it reframed the conversation from “Can AI automate jobs?” to “What do workers want AI to automate or augment?”
The Top Three Findings
1. Workers are open to automation, but on their terms. Workers are open to AI automating about 46% of tasks, mainly to free up time for more meaningful or high‑value work, not to eliminate their roles.
2. People prefer partnership, not replacement. Stanford introduced the Human Agency Scale (H1 to H5), measuring the level of human involvement workers prefer. About 45% of workers prefer equal partnership (H3) with AI, not full automation.
3. Tasks fall into four zones based on AI capability and worker desire:
-
Green Light: High desire + high capability (ideal for rollout). Examples: Tax preparers scheduling appointments or timekeeping clerks adjusting payroll.
-
Red Light: Low desire + high capability (risk of backlash). Examples: Editors writing stories or media managers reviewing visual content.
-
R&D Opportunity: High desire + low capability (unmet innovation). Examples: Financial analysts coordinating investment prep or quality control managers reviewing complex data.
-
Low Priority: Low desire + low capability (minimal impact). Examples: Ticket agents tracing lost baggage or art directors presenting layouts.
My Key Concern: Why So Much Focus on the Red Zone?
What surprised me is how many of the AI tools are targeting tasks that AI can do, but that workers don’t want it to do. Why?
-
Tech & Product Bias: Investors and startups chase quick wins in areas like coding automation or data handling. They’re easier to automate, even if they fall in the Red Zone.
-
Lack of Worker-Centered Data: Until this study, before WORKBank, no one systematically captured what workers actually want. Most roadmaps have been tech-driven, not worker‑driven (or what really end-users need).
-
Path Dependence: Startups and researchers often iterate in familiar technical domains, leaving high‑desire but hard-to-solve problems (R&D Opportunity) out of the game.
What does this mean for Your Strategy?
Here’s how to approach each zone strategically.
Why This Matters
Many still think about AI as a replacement. But this study shows that People want AI as collaborators, not replacements. They’re looking for tools that take the grunt work off their plate so that they can focus on strategy, nuance, and creativity.
The issue is that 41% of AI deployments today are focused on tasks where workers don’t want AI, and this misalignment leads to friction, frustration, and failed adoption.
41% of AI deployments are focused on tasks where workers don’t want automation.
The Executive Takeaway:
-
Recalibrate your AI strategy through your people’s lens (at least as a starting point). You can use frameworks like WORKBank and HAS, referenced in this article.
-
Shift your investment focus toward high‑desire, underserved areas (e.g., operational coordination or planning tools).
-
Double down on upskilling soft skills, such as communication, emotional intelligence, judgment, and leadership.
-
Involve employees early in pilot design. Build with them, not just for them, especially in the Red Zone.
If you’re seeing pushback on an AI rollout or if adoption is lagging, the issue might not be the tech. It might be that you’re solving the wrong problem or a task that falls in the Red Zone.
Start treating worker sentiment as part of your ROI model.
-
Reference: Future of Work with AI Agents: Auditing Automation and Augmentation Potential across the U.S. Workforce, Yijia Shao, Humishka Zope, Yucheng Jiang, Jiaxin Pei, David Nguyen, Erik Brynjolfsson, Diyi Yang. Published by arXiv:2506.06576.
If you’re navigating this shift in your organization or just curious about what’s next, I’d love to connect. I’m always interested in exchanging ideas, learning from others, and helping leaders think through the road ahead.
🤝 And if you find this article helpful, feel free to share it with your network.