Study suggests human jobs safe from AI threat, for now

Despite the pervasive apprehension about AI potentially displacing jobs, a Massachusetts Institute of Technology (MIT) study, suggests that at present, AI is economically impractical for most roles.

The study, titled “Beyond AI Exposure,” delves into the feasibility of replacing human labor with artificial intelligence (AI) across 800 occupations in the United States, examining 1,000 visually assisted tasks. The investigation specifically focused on jobs utilizing computer vision, such as teachers, bakers, and property appraisers.

The findings reveal that merely 23% of worker wages could be efficiently substituted by AI, emphasizing the financial constraints associated with widespread AI adoption.

Even with a projected 20% annual reduction in costs, the study anticipates that decades are needed for computer vision tasks to become financially viable for companies.

For instance, in a hypothetical bakery scenario outlined in the study, computer vision was employed to scrutinize ingredients for quality control. However, this specific task constitutes only six percent of their overall work and would incur higher installation and operational costs for the technology compared to a human performing the same task.

Financed by the MIT-IBM Watson AI Lab, the research employed online surveys to assess tasks in 800 occupations. Notably, the study challenges the prevailing notion that AI will rapidly replace human jobs, stating, “Machines will steal our jobs” is a sentiment often voiced during technological upheavals.

The researchers argue that only a fraction of worker compensation exposed to AI computer vision—23%—is economically justifiable for automation due to substantial upfront costs.

Despite concerns raised by the advent of large language models like OpenAI’s ChatGPT, the study underscores the current impracticality of widespread AI adoption in the workforce.

OpenAI’s CEO, Sam Altman, further asserted in Davos that artificial general intelligence (AGI) would have a more modest impact on the world and job landscape than commonly feared.