Human workers still have one crucial advantage over AI – but you might not be happy to hear what it is

Key Takeaways:

– A new study by MIT suggests that concerns about AI replacing human workers in the near future are unwarranted because humans are cheaper to use.
– The study, funded by the MIT-IBM Watson AI Lab, focuses on implementing AI systems in industries that require computer vision skills.
– Even with generous predictions of AI cost reductions, the study estimates that it would still take several years for AI to become a viable option to replace humans.
– The report suggests that by 2026, only 50% of vision tasks would have a machine economic advantage, even with a high 50% annual cost decrease for AI systems.
– With a more conservative 20% annual cost reduction, it would take “decades” before replacing computer vision task human workers with AI would be economically viable.
– The study highlights AI’s reliance on statistical and symbolic reasoning compared to the broader range of intelligence covered by human workers.
– While some sectors like banking, marketing, healthcare, legal, retail, and transportation could benefit from automation, a Goldman Sachs study suggests that only around 18% of the global workforce could be displaced by generative AI.
– The same study also suggests that generative AI could raise annual US labor productivity growth by just under 1.5pp over a 10-year period following widespread adoption.
– The long period over which a small boost in productivity is expected provides some confidence in the future of human jobs.
– Overall, the scale of AI’s effects on the world is yet to be seen, but the studies suggest that human jobs will remain secure for several more decades.

TechRadar:

According to a new study conducted by researchers are the Massachusetts Institute of Technology (MIT) (via SiliconAngle), increasing concern that AI could replace human workers imminently is unwarranted because humans are cheaper to use.

The “Beyond AI Exposure” study, financed by the MIT-IBM Watson AI Lab, delves into the practicalities of implementing AI systems in various industries, focusing gon tasks requiring computer vision skills.

Source link

AI Eclipse TLDR:

A new study conducted by researchers at the Massachusetts Institute of Technology (MIT) suggests that concerns about AI replacing human workers in the near future are unfounded. The study, called “Beyond AI Exposure” and financed by the MIT-IBM Watson AI Lab, focuses on the implementation of AI systems in various industries, specifically tasks requiring computer vision skills.

The researchers at MIT analyzed the cost reductions of AI and found that even with generous predictions, it would still take several years for AI to become a viable option to replace humans. The extent of AI’s capability to replace human workers would also be very limited.

The report estimates that, even with a high 50% annual cost decrease, it would not be until 2026 that 50% of vision tasks could be economically advantageous for machines. The researchers also noted that even by 2042, there would still be tasks where human labor has an advantage over computer vision.

The analysis attributes this to the fact that impressive analysis and recognition tasks can only be performed by immensely powerful and costly AI systems. Furthermore, the study highlights that human intelligence covers a much larger area, including conscious and subconscious thoughts, while AI relies on statistical and symbolic reasoning.

While sectors such as banking, marketing, healthcare, legal, retail, and transportation could benefit from some automation, a separate Goldman Sachs study revealed that around 18% of the global workforce could be displaced or replaced by generative AI. However, the same study also suggests that generative AI could raise annual US labor productivity growth by just under 1.5% over a 10-year period following widespread adoption.

Overall, the MIT study indicates that AI’s impact on the world is still uncertain, but it is unlikely to replace human workers for several more decades.