A future in which autonomous systems “outperform humans at the most economically valuable tasks,” as described in OpenAI's mission statement, sounds like a hellscape to MIT economics professor David Auter. .
A world in which humans provide only a generalized, undifferentiated flow of labor and wealth to the owners and rights holders of AI systems will be like Wall-E meets Mad Max. , he says.
But it doesn't have to be that way. In a paper published through the National Bureau of Economic Research, titled “Applying AI to Rebuild Middle-Class Jobs,'' Auter argues that concerns about a future in which humans will be rendered powerless by AI are misplaced. Yes, and in fact, they argue that AI can improve many jobs. of the middle class.
Elon Musk's prediction in a recent interview with British Prime Minister Rishi Sunak that there will come a time when we won't need a job, and AI pioneer Jeffrey Hinton's recommendation to “get a plumbing job” Citing advice, Autor argues for the future. You will never run out of work. He argues that falling birth rates and a shrinking workforce will ensure a labor shortage.
The question centers on what the available jobs will be. Auter bases his argument on the harms of the Information Age, where the emergence of AI as a support tool has devalued the procedural expertise of middle-class workers and shifted power to elite decision-makers. I believe it provides a path back.
“The unique opportunity that AI offers humanity is to resist the processes initiated by computerization and expand the relevance, scope, and value of human expertise to more workers,” he wrote. There is.
“Artificial intelligence can weave information and rules based on acquired experience to support decision-making, so more workers with the necessary basic training are now available to elite professionals such as doctors. be able to perform high-stakes decision-making tasks for lawyers, software engineers, and university professors.”
If you are a highly paid professional in an occupation with limited qualifications, this may not seem like an ideal outcome. But there is precedent for such a change.
For example, Auter points to the work of registered nurses (RNs), who earn additional master's degrees that certify them to perform tests and manage services traditionally reserved for doctors.
The number of nurses in the United States nearly tripled to about 224,000 between 2011 and 2022, he said, and that number is expected to grow 40 percent over the next 10 years. What made it possible? Beyond decisions by medical professionals in the 1960s to better utilize the skills of registered nurses and change medical regulations, Auter points to information technology, particularly electronic health records.
“Improved electronic medical records and communication tools are enabling NPs to make better decisions,” Auter wrote, and AI similarly empowers others to make decisions that would otherwise be left to experts. claims that it can be given to employees of
He points to several studies on the impact of GitHub's Copilot and OpenAI's ChatGPT on computer programming and writing tasks, respectively. Neither eliminated the need for specialized knowledge, but both helped increase the productivity of fewer workers.
“Artificial intelligence is a technology that reverses this,” Auter argues. “By providing decision support in the form of real-time guidance and guardrails, AI can complement some of the riskier decision-making tasks currently assigned to elite professionals such as doctors, lawyers, and programmers. It has the potential to enable more workers with knowledge of the field and educators to perform.”
He says that just as the Industrial Revolution made consumer goods more affordable, AI will improve the quality of jobs for people without degrees, reduce income inequality, and improve health care, education, It is said that the cost of legal consultation can be reduced.
Autor has made it clear that he does not expect AI to eliminate the need for specialized knowledge. He says you can't have untrained people do skilled tasks like inserting catheters. However, workers who have some basic knowledge of the task can level up.
Autor says this outcome is not inevitable. “But it is technically plausible, economically consistent, and morally compelling,” he concludes. “Recognizing this potential, we should ask not what AI can do for us, but what we want it to do for us.” ®
