A job that AI can do – and it shouldn’t be

Unlock Editor’s Digest Lock for Free

FT editor Roula Khalaf will select your favorite stories in this weekly newsletter.

Generator AI is a transformative technology with the potential to redefine the nature of work. Understanding the role in the workplace, and what is different from past automation requires a transition from what AI can do to what it should.

A typical analysis of Genai’s impact on workers focuses on whether the technology is capable of performing a particular job. Such studies often disrupt work and assess the share of components that technology can implement. For example, common tasks for call center customer service representatives include interacting with customers, recording interactions, resolving or escalating concerns. Genai can handle these tasks, meaning that it could replace such workers.

However, consider emergency services telephone operators, which are occupations that initially appear to be comparable. The two jobs share many similar tasks. Should we expect them to face the same level of automation risk? The answer is more subtle than just technical capabilities. Beyond ethical considerations, automating these roles introduces complex trade-offs, including economics, task design and operational interdependence.

When considering automation, organizations believe that four important questions should be considered:

First of all, how complicated is the task? Complexity is a key driver of both the human labor and the cost of AI. Emergency Service Deployers solve a variety of issues, including levels of complexity that outweigh the repetitive interactions of customer service personnel. In general, the more complicated the task, the less likely it is to be automated, as humans are – for now they are better than machines with increased complexity.

Secondly, how often do tasks take place? The higher the frequency, the more likely it is to be automated. The machine has distinct advantages in maintaining speed over the long term. Frequently repeated client interactions enhance the economic case for customer service personnel’s AI exchange.

Third, how interconnected are the tasks? In providing services or creating products, many jobs are involved in a chain of interconnected tasks that are often completed by various workers and machines. What happens during handoffs between tasks is often overlooked. Fragmentation costs arise from inefficiencies and errors in the handoff process.

The first task for a customer service representative will involve conversations with the customer, but the final task is to solve the problem. Handoffs between these tasks can be costly when different workers or machines are involved. If the workers processing the final solution do not initially interact with the customer, they will need additional time to see all previously collected information.

High fragmentation costs should discourage companies from splitting tasks between humans and generated AI, even if they are technically viable. Automating your initial triage calls with emergency services may seem cost-effective, but important information can be lost during the transition from AI to human dispatchers.

Fourth, how much does the cost of failure cost when performing a task? Mistakes caused by emergency dispatchers pose a significant risk, especially in life and death situations. Also, genai is less accurate than past forms of automation.

These questions should help you to consider automation and guide the company and explain why Genai is affecting a particular occupation more than other occupations. For example, consider a computer programmer. With extensive and well-documented coding examples, Genai can provide effective solutions for complex tasks. The high frequency and repetition of many coding tasks are perfect for genai.

Long before Genai, programmers split large coding projects, and innovations such as distributed development platforms and modular design reduced fragmentation costs. A secure test environment can detect many errors in code written by genai at a low cost, thus keeping the cost of failures low. Within our framework, these features help explain why programmers who are traditional automation beneficiaries are facing an increasing number of confusion from Genai.

The four questions above highlight why generating AI is unique as an automation technology. As it evolves, Genai demonstrates its ability to manage fast and complex tasks, making it more versatile than traditional automation. By providing a seamless interface and natural language processing capabilities, Genai gradually reduces fragmentation costs compared to traditional automation. However, the uncertainty surrounding Genai output can potentially increase the risk of task failure.

Generator AI is a transformative technology with the potential to restructure the labor market. Its ultimate impact and potential for adoption are shaped by the structure of tasks within a particular occupation. Task complexity, frequency, fragmentation costs, and failure costs affect the overall cost reduction and hidden costs.