← Back to Blogs

Rethinking Automation, Augmentation, and Human Agency

As AI agents become more capable and integrated into the workplace, the debate over “automation vs. augmentation” is intensifying. A new paper, "Future of Work with AI Agents: Auditing Automation and Augmentation Potential across the U.S. Workforce", from Stanford University, delivers the most comprehensive look yet at how American workers actually want to work with AI, and what tasks are ready (or not) for handoff to intelligent systems.

Beyond the Automation Hype: A Worker-Centric Approach

Most AI research focuses on technical benchmarks or narrow job categories. But this study flips the script. Rather than only asking what AI can do, the authors audit what workers themselves want AI to do for them. Using an innovative, audio-enhanced survey and a novel Human Agency Scale (HAS), they collected nuanced insights from 1,500 domain experts across 104 occupations—building a new public database called WORKBank.

What Did They Find?

1. Workers Want Relief From Repetition, Not Replacement

2. The Four Zones of AI Task Adoption

The paper maps every task to one of four “zones,” based on worker desire and current AI capability:

Surprisingly, much current investment clusters in low-desire areas, not the green-light or opportunity zones.

3. Human Agency Matters More Than Ever

The Human Agency Scale (H1 H5) quantifies how much human involvement is desired (from fully automated to essential human control). The key finding? Most workers and experts don’t want to fully “outsource” their jobs to AI. Instead, they envision collaboration—equal partnerships where AI augments their skills, rather than replacing them.

4. The Shift in Skills: From Data to Human Connection

As AI eats more information-processing work, what’s left for humans? According to the data, interpersonal, creative, and organizational skills are rising in value—while pure data wrangling and routine tasks drop.

Why Does This Matter?

Bottom Line

This landmark study marks a new era in the AI-at-work conversation—one that centers human desires and the future of agency, not just efficiency. The message is clear: AI should empower workers, not erase them.

Curious to see the full WORKBank data or read the paper? Check it out on arXiv.