The Future of Work: What AI Should Do vs. What It Can Do

Generative AI is a transformative technology poised to redefine the nature of work. To understand its role effectively, we need to shift our focus from what AI can do to what it should do.

Assessing the Impact of Generative AI

Typical evaluations of GenAI’s impact on employment often examine whether the technology can perform specific jobs, breaking down roles into tasks that AI can execute. For example, a customer service representative may have tasks such as interacting with customers and resolving concerns, which GenAI can handle, suggesting potential displacement.

However, consider an emergency service phone operator—though the tasks may seem similar, the risk of automation is more nuanced. Factors beyond mere technical capability, including ethical considerations, economics, and operational interdependencies, come into play.

Key Questions for Organizations

Organizations should contemplate four pivotal questions regarding automation:

1. How complex is the task? Complexity drives both human labor and AI costs. Tasks like those handled by emergency service dispatchers are complex and less likely to be automated, while repetitive tasks like those of customer service representatives are more susceptible to automation.
2. How frequent is the task? Tasks performed frequently are more likely to be automated, as machines excel in maintaining speed over prolonged periods.
3. How interconnected are the tasks? Many jobs involve a chain of interconnected tasks. High fragmentation costs can arise from inefficiencies in handoffs between tasks, discouraging companies from splitting tasks between humans and AI.
4. What is the cost of failure? In high-stakes environments like emergency services, the cost of mistakes is significant. GenAI may not always match the precision required in these scenarios.

The Unique Nature of Generative AI

These considerations explain why GenAI impacts certain occupations, like computer programming, more than others. Programmers face increased disruption due to the repetitive and well-documented nature of many coding tasks, making them a good fit for GenAI.

As GenAI continues to evolve, its ability to manage complex tasks at high speed makes it more versatile than traditional automation. It lowers fragmentation costs, but the uncertainty of its outputs may increase the risk of failure.

Generative AI’s transformative potential could reshape labor markets significantly, influenced by task structures within occupations. The complexity, frequency, fragmentation costs, and the cost of failure together impact the balance between realized cost savings and hidden costs.