There have been plenty of warnings about job losses due to AI, particularly in the world of IT and in the reduction of entry-level positions. Doom mongers’ claims that AI is going to eradicate all our jobs look to be exaggerated but there is little room for complacency as there are some roles most definitely at risk — and software engineers, in particular, should be looking over their shoulders.
Economists Luis Garicano, Jin Li, and Yanhui Wu, report in a new research paper that jobs that can be unbundled — with tasks AI can perform easily separated from those requiring human input — are at greater risk than jobs where the two kinds of tasks are indissociable. “The effect of AI depends on how costly it is to break the bundle,” they wrote. This means that “In strong-bundle occupations, where tasks are not independently reallocable, AI improves performance inside the job, but does not remove the human from the bundle,” consequently, protecting jobs.
Garicano et al don’t provide examples of such jobs, but point out that jobs cannot be unbundled when there is a high coordination cost to separating their tasks, such as when one person must assume liability for all the tasks, when performing one task improves performance at another, or when a shared context is needed to perform all the tasks.