The works that AI can do, and those who should not

Unlock the editor’s summary for free

Roula Khalaf, editor of the FT, selects her favorite stories in this weekly newsletter.

Generative AI is a transformative technology that has the potential to redefine the nature of work. Understanding its role in the workplace and what makes it different from past automation requires a change of what ai can do to what ought do.

Typical analysis of Genai’s impact on workers focus on whether technology can perform specific jobs. These studies often break down a job and evaluate the proportion of the constituent tasks that technology can execute. For example, common tasks for a customer service representative in a call center include interacting with customers, registering interactions and solving or growing concerns. Genai can handle these tasks, which implies that he could move to such workers.

But consider an occupation that initially may seem equivalent: an emergency telephone operator. The two works share many similar tasks. Should we expect them to face equal levels of automation risk? The answer is more nuanced than the technical capacity alone. Beyond ethical considerations, the automation of such roles introduces complex complexes that involve economy, task design and operational interdependence.

We believe that organizations should consider four fundamental questions when contemplating automation.

First, how complex is the task? Complexity is a key driver of the costs of human work and AI. Emergency service dispatators solve a wide variety of problems, which implies a level of complexity that exceeds the repetitive interactions of a customer service representative. In general, the more complex the task is, the less likely it is automated, since humans are, for now, better than machines to handle greater complexity.

Second, how frequent is the task? The higher the frequency, the more likely it will be automated. Machines have a clear advantage to maintain speed for prolonged periods. Interactions frequently repeated with customers strengthen the economic case for the replacement of the customer service representatives.

Third, how interconnected are the tasks? By providing a service or creating a product, many works participate in a chain of interconnected tasks often completed by different workers and machines. What happens during the transfer between tasks is often overlooked. Fragmentation costs arise from inefficiencies and errors in the transfer process.

The initial task for a customer service representative implies talking with the client, while the final task is to solve your problem. When there are different workers or machines involved, the transfer between these tasks can be expensive. If the worker who handles the final resolution did not initially interact with the client, additional time would be needed to review all the information previously collected.

High fragmentation costs should discourage companies to divide tasks between humans and generative AI, even if it is technically feasible. The automation of the initial triage call in emergency services may seem profitable, but crucial information could be lost during the transition from AI to a human dispatcher.

Fourth, when executing a task, what is the cost of the fault? The errors of emergency dispatators pose significant risks, particularly in life or death situations. And Genai may be less accurate that some past forms of automation.

These questions should guide companies that consider automation and help explain why Genai affects certain more than others. Consider computer programmers, for example. Examples of extensive and well -documented coding allow Genai to provide effective solutions even for complex tasks. The high frequency and repetitiveness of many coding tasks fit well with Genai.

Long before Genai, programmers divided large coding projects, and innovations such as distributed development platforms and modular design have reduced fragmentation costs. The safe test environments maintain the low failure cost, since many errors in the code produced by Genai can be detected economically. Within our framework, these characteristics help explain why programmers, traditionally beneficiaries of automation, face a greater interruption of Genai.

The four previous questions highlight which makes generative AI unique as automation technology. As it evolves, Genai is demonstrating its ability to manage complex tasks at high speed, which makes it more versatile than traditional automation. By offering a perfect interface and natural language processing capabilities, Genai progressively reduces fragmentation costs compared to traditional automation. However, the uncertainty surrounding Genai production potentially increases the risk of failure in a task.

Generative AI is a transformative technology with the potential to remodel labor markets. Its final impact and its probability of adoption are formed by the structure of the tasks within a particular occupation. The complexity of the tasks, their frequency, fragmentation costs and the cost of the failure, taken together, influence the balance between manifest cost savings and hidden costs.