The Dark Ages were not entirely dark. Advances in agriculture and building technology increased medieval wealth and caused a wave of cathedral building in Europe. But it was a time of deep inequality. The elite got virtually all the economic benefits. In England, while Canterbury Cathedral soared, between 1100 and his 1300 the wealth of the peasants did not increase at all. Life expectancy he hovered around 25 years. Chronic malnutrition was widespread.
“We have long struggled with shared prosperity,” says Simon Johnson, a professor at the Massachusetts Institute of Technology. “Every cathedral my parents took me to see in Europe is a symbol of despair and deprivation made possible by high productivity.”
At first glance, this may seem irrelevant to life in 2023. But Johnson and his MIT colleague economist Daron Acemoglu think there is a connection. Technology drives economic progress. As innovation takes hold, the eternal question is, “Who will benefit?”
Scholars believe this also applies to automation and artificial intelligence, the focus of Acemoglu and Johnson’s new book, Power and Progress: Our Thousand-Year Struggle for Technology and Prosperity, published this week in PublicAffairs. In it, they consider who will benefit from past innovations and who will benefit economically and politically from today’s AI.
“This book is about the choices we make when it comes to technology,” Johnson said. “It’s a very MIT theme.
AI could evolve into a useful force, Johnson said. However, “Many algorithms are designed to replace humans as much as possible. We need to make it work for humans, not replace it.There was automation before, but it needed new tasks for humans to do and enough power in society.”
According to Acemoglu and Johnson, today AI is both a tool of social control for some governments and at the same time bringing wealth to a minority. “AI’s current path is neither good for the economy nor good for democracy, and unfortunately these two problems reinforce each other,” they write.
A Return to Shared Prosperity?
Acemoglu and Johnson have collaborated before. In the early 2000s, he published influential papers on political and economic progress with political scientist James Robinson. A research professor at the Massachusetts Institute of Technology, Acemoglu is the author of a book on political institutions and growth, Why States Fail (2012), and The Narrow Corridor, which casts freedom as an outcome that is never guaranteed. 2019) with Robinson. of social struggle.
A Ronald A. Kurtz Professor of Entrepreneurship at MIT Sloan School of Management, Johnson is the author of 13 Bankers (2010) on financial reform and co-author with MIT economist Jonathan Gruber, Jump Start America” (2019). Call for more investment in scientific research.
In Power and Progress, the authors emphasize that technology has produced significant long-term benefits. They wrote that “we are much richer than our ancestors” and that “advances in science and technology are an important part of that story.”
Yet much suffering and oppression occurred over the long term, not just in the Middle Ages.
“It’s been a 100-year struggle of the Industrial Revolution for workers to get some of the massive productivity gains in the textile and railroad industries,” says Johnson. Broader progress came through increased labor force and elected government. When the U.S. economy enjoyed her 30 years of impressive growth after World War II, its gains were widely distributed, but not recently.
“We are proposing that we can get back on that track of shared prosperity, repurposed technology for everyone, and increased productivity,” says Johnson. “We had it all after the war. We can take it back, but in the current form of our obsession with machine intelligence, we cannot. we are thinking.”
Calls for “usefulness of machines” instead of “somewhat automated”
What do Acemoglu and Johnson think AI is missing? First, they believe that AI development is too focused on mimicking human intelligence. Academics are skeptical of the idea that AI reflects all of human thought, even things like the chess program AlphaZero are specialized sets of instructions.
Or, for example, an image recognition program — is it a husky or a wolf? — uses a large dataset of past human decisions to build a predictive model. But these often rely on correlations (likely having a husky in front of your house) and can’t replicate the same cues that humans rely on. Researchers know this, of course, and continue to improve their tools. But Acemoglu and Robinson argue that even though AI is designed to replace human jobs, many AI programs are less agile than the human mind and are not the best replacements. .
Acemoglu, who has published many papers on automation and robotics, calls these alternative tools “so-so technology.” Supermarket self-checkout does not provide meaningful economic productivity. Just transfer jobs to customers and wealth to shareholders. Or, even among more advanced AI tools, for example, a customer service line using AI that doesn’t address a specific issue can frustrate people, complain after actually contacting a human, and reduce the efficiency of the entire process. There is a nature.
Overall, Acemoglu and Johnson write, “Neither traditional digital technologies nor AI can perform important tasks involving social interaction, adaptation, flexibility and communication.”
Rather, growth-minded economists favor technologies that create “marginal productivity” increases that force firms to employ more workers. Rather than aiming to eliminate medical professionals such as radiologists, where AI development has been much anticipated but has not materialized, Acemoglu and Johnson are expanding what home care workers can do with AI tools. It suggests that it may be possible to increase the value of the service without reducing the number of home healthcare workers. .
“We believe there is a fork in the road, but it’s never too late. AI is a great opportunity to reassess the utility of machines as a design philosophy,” says Johnson. increase. “And to find ways to put the tools in the hands of workers, including low-wage workers.”
Definition of discussion
Another AI problem Acemoglu and Johnson are concerned about is surveillance technology, facial recognition tools, intensive data collection, and misinformation spread by AI.
While severely restricting freedom of expression, China employs AI to create a “social credit” score for its citizens and conducts heavy surveillance. Elsewhere, social media platforms use algorithms to influence what users see. By emphasizing “engagement” over other priorities, you can spread harmful misinformation.
Indeed, through “power and progress,” Acemoglu and Johnson argue that the use of AI enables those economically benefiting from gaining political influence and power at the expense of broader democratic participation. It emphasizes that you can build a self-reinforcing dynamic of being able to.
To change this trajectory, Acemoglu and Johnson advocate a broad menu of policy responses, including Internet users’ data ownership (an idea of technologist Jaron Lanier). Tax reform that rewards employment over automation. Government support for diverse high-tech research directions. Repeals Section 230 of the Communications Decency Act of 1996, which protects online platforms from regulation and legal action based on the content they host. and a digital advertising tax (aimed at limiting monetization of algorithmic misinformation).
Johnson believes people of all ideologies have incentives to support such policies, and “we are not advocating partisan arguments,” he says.
Other scholars praise “strength and progress”. Michael Sandel, Ann T. Bass and Robert M. Bass Professor of Public Administration at Harvard University, called the book “showing how technology can be channeled to advance the public good.” It is a book full of hope and hope,” and a “must read book.” For all concerned about the fate of democracy in the digital age. “
Acemoglu and Johnson want to broaden the public debate on AI beyond industry leaders, abandon notions of AI’s inevitability, and rethink human agency, social priorities, and economic potential. increase.
“Discussions about new technologies should center not only on how great new products and algorithms are, but whether they work for people or against them,” they write.
“We need these discussions,” Johnson said. “There is nothing specific to technology. can do.”
