-->

⏱️ Estimated Reading Time: 12 minutes

The AI Era: What We Really Want to Know is the ‘Voice of Workers’

With the explosive growth of generative AI, AI agents are deeply penetrating our daily lives and work. However, most research so far has focused on “What can AI do?” while neglecting the fundamental question of “What do workers actually want?”

The latest research published by Stanford University researchers fills this gap through a groundbreaking attempt. Through a large-scale survey of 1,500 field workers and 52 AI experts, they systematically analyzed labor market changes in the age of AI agents from the workers’ perspective.

WORKBank: A Worker-Centered AI Audit Framework

Innovative Research Methodology

The core of this research is the construction of a new database called WORKBank. Based on the U.S. Department of Labor’s O*NET occupational database, they conducted dual evaluations of 844 specific tasks across 104 occupations, examining automation preferences and reasons from workers’ perspectives, and automation feasibility levels achievable with current AI technology from experts’ perspectives.

Human Agency Scale: The Spectrum of Collaboration

Moving beyond the simple “automatable/non-automatable” dichotomy, the researchers developed a five-level scale called the Human Agency Scale. This scale ranges from complete automation where AI performs tasks independently, to AI-led collaboration where AI leads but humans review, equal partnership where humans and AI collaborate as equals, human-led collaboration where humans lead with AI assistance, and human-essential tasks that must be performed by humans.

Surprising Research Results: The Gap Between Expectations and Reality

Worker Acceptance of Automation

The research found that workers positively accepted automation in 46.1% of tasks. The main reason was the desire to “automate repetitive and low-value tasks to focus on high-value work.” This shows stronger strategic collaboration intentions rather than vague fears about AI.

The Severity of Demand-Technology Mismatch

The researchers analyzed worker automation preferences and AI technology levels in a matrix format, deriving four zones: the Green Light Zone where both demand and technology are ready, the Red Light Zone where technology exists but workers don’t want it, the R&D Opportunity Zone where demand is high but technology is lacking, and the Low Priority Zone where both demand and technology are low.

The shocking discovery was that a significant portion of current startup investments are concentrated in the Red Light Zone and Low Priority Zone. This shows how disconnected technology-centered approaches are from actual worker demands.

Preferences and Reality of Collaboration

45.2% of workers preferred equal partnership with AI. In contrast, AI experts judged that lower levels of automation were technically feasible. This perception gap predicts considerable friction in future AI adoption processes.

The Great Transformation of Future Job Competencies

The research discovered another important trend: the shift in core job competencies. Competencies decreasing in importance include data analysis and information processing, repetitive computational tasks, and simple document writing. Competencies increasing in importance include interpersonal relationships and communication, organizational operation and coordination abilities, creative problem solving, and ethical judgment and decision-making.

Insights: New Strategies for the AI Era

Implications for Investors and Companies

Current AI investment patterns are trapped in technology-first thinking. Real market opportunities lie in the R&D Opportunity Zone. Investment should focus on areas that workers desperately want but where technology is not yet complete.

Conditions for successful AI startups include technological excellence combined with worker demand alignment, human-AI collaboration models rather than complete replacement, and win-win structures through improved work efficiency.

Recommendations for Policymakers

Fundamental redesign of education and retraining policies is needed, shifting from technology-centered education to human-centered competency education, continuous skill upgrades through lifelong learning systems, and strengthening AI literacy and collaboration capabilities.

Individual Preparation Strategies

Conditions for professionals who will survive in the future include the ability to use AI as a tool, ethical judgment in complex situations, communication and coordination abilities with various stakeholders, and creative thinking and innovation capabilities.

Future Outlook: The Age of Collaboration is Coming

Short-term Outlook (1-3 years)

We can expect rapid automation of Green Light Zone tasks, widening productivity gaps between professionals familiar with AI tools and those who are not, and increasing need for standardization of AI collaboration processes within companies.

Medium-term Outlook (3-5 years)

This period will see technological breakthroughs in the R&D Opportunity Zone, formation of worker-centered AI service ecosystems, and full-scale work redesign considering the Human Agency Scale.

Long-term Outlook (5-10 years)

Human-AI collaboration will establish as the standard work method, new job categories will emerge including AI collaboration specialists and ethical AI supervisors, and education systems centered on soft skills will be fully established.

Conclusion: Humans Must Be at the Center, Not Technology

The most important message this research conveys is that “the direction of AI development should be determined by humans, not technology.” Until now, we have focused only on “What can AI do?” But the real question is “How can AI make human life richer?”

The WORKBank research shows that this human-centered approach is not just an ideal but a concrete and feasible methodology. We must listen to workers’ voices, find the balance point between technology and demand, and create a path for AI development that preserves human dignity.

The winner in the AI era will not be the one with the most sophisticated algorithms, but the one who creates an ecosystem where humans and AI grow together.

References:

  • Original Paper - “Auditing AI Agents for Job Integration: A Framework for Assessing Worker-Oriented Automation Preferences”
  • Stanford University HAI (Human-Centered AI Institute)
  • O*NET Occupational Database

참고