(Illustration by Hugo Herrera)
In November 2022, OpenAI grabbed headlines by releasing ChatGPT, and “generative AI” became a term used in households around the world. According to Open AI CEO Sam Altman, “A lot of people working on AI pretend that it’s only going to be good; it’s only going to be a supplement; no one is ever going to be replaced.” But, he adds, “jobs are definitely going to go away, full stop.” Two major labor strikes this year, by the Writers Guild of America and the actors union (SAG-AFTRA), have helped focus attention on the threat AI poses to well-paid jobs.
Artificial Intelligence is poised to upend work around the world over the coming decades. Whether this lessens or increases inequality remains to be seen. AI technology is relatively new, but the impacts of previous high-impact innovations like the power loom, steam engines, electricity, and digital computers shed ample light on what could happen next. The consequences of any technology depend on who gets to make pivotal decisions about how the technology develops. This is doubly true for AI, because these new tools can be developed for many different types of activities, with the potential to spread rapidly in every sector of the economy and in every aspect of our lives.
Too many commentators see the path of technology as inevitable. But the historical record is clear: technologies develop according to the vision and choices of those in positions of power. As we document in Power and Progress: Our 1,000-Year Struggle over Technology and Prosperity, when these choices are left entirely in the hands of a small elite, you should expect that group to receive most of the benefits, while everyone else bears the costs—potentially for a long time.
Rapid advances in AI threaten to eliminate many jobs, and not just those of writers and actors. Jobs with routine elements, such as in regulatory compliance or clerical work, and those that involve simple data collection, data summary, and writing tasks are likely to disappear.
But there are still two distinct paths that this AI revolution could take. One is the path of automation, based on the idea that AI’s role is to perform tasks as well as or better than people. Currently, this vision dominates in the US tech sector, where Microsoft and Google (and their ecosystems) are cranking hard to create new AI applications that can take over as many human tasks as possible.
The negative impact on people along the “just automate” path is easy to predict from prior waves of digital technologies and robotics. It was these earlier forms of automation that contributed to the decline of American manufacturing employment and the huge increase in inequality over the last four decades. If AI intensifies automation, we are very likely to get more of the same—a gap between capital and labor, more inequality between the professional class and the rest of the workers, and fewer good jobs in the economy.
There is a second, very different path available to us, however. This path would focus on creating new tasks and capabilities for humans, rather than sidelining them. This too is not new in history. The augmentation of human capabilities, by creating new tasks and providing better tools and information for workers, was the bedrock of wage growth and shared prosperity during the decades that followed World War II.
In principle, AI could magnify the possibilities for human-complementary technological change. It could enable the development of a wide array of tools that provide better information to human decision-makers. Critically, this need not be limited to office workers and professionals. Much of the workforce today, from blue-collar workers in factories to electricians, plumbers, educators, and health care providers, depends on problem-solving and real-time decision-making. With better context-specific information, these workers could become more productive in the tasks they are performing and venture into new, more complex tasks.
Alas, this more hopeful path is not where we are heading. Three big social changes would be necessary for such a path, and each one of them is a tall order.
First, management needs to see and understand workers as a key resource whose productivity should be augmented, whose information should be improved, and whose training should be a priority. But the dominant perspective in most C-suites views labor as a cost to be cut, either to withstand competition or to better remunerate shareholders. What gets lost in the rush to reengineer the corporation with fewer workers is the long-term health of the companies, which never achieve the promised productivity gains from so-so automation and offshoring. Companies must recognize that, in reality, labor is a critical resource for productivity growth.
Second, the tech sector needs to prioritize helping workers, rather than focusing on tools of automation and surveillance. The industry’s vision has too long been shaped by an ill-advised quest for artificial general intelligence and autonomous machine intelligence, and many technologists are still preoccupied by showing how their algorithms can reach “human parity.” In practice, that means automating as many tasks as possible.
Third, labor needs to have a voice in how technologies are used. This voice is critical not only for resisting excessive emphasis on labor cost-cutting and automation. It is also essential because workers typically know which parts of their jobs would benefit from automation and which would not. They recognize which tasks could be made more efficient, freeing them to spend time on more productive activities, or even creating new opportunities for increased productivity. Increased worker buy-in reduces businesses’ incentives for further intensification of worker monitoring and surveillance. It also ensures that any productivity gains are shared more fairly between capital and labor.
These three social changes are possible, even if very unlikely to happen without a coordinated effort.
There is much greater pushback against the “shareholder values revolution”—which elevated cost-cutting and the interests of shareholders at the expense of workers—than at any other time over the last four decades. A new philosophy prioritizing long-term productivity growth may yet emerge, which management and workers can work together to achieve.
The tech sector can change, too. It was always an aberration that arguably the most powerful industry in the US (and the world) would ignore all social responsibility and elevate the virtues of “disruption” with an almost religious zeal. In its next, more mature phase, the industry could focus on providing better tools for workers and augmenting human capabilities, rather than replacing them wholesale.
And labor can become better organized and better focused on charting a new course for involvement in the production process during the age of AI, which will also enhance productivity. In fact, pro-union sentiment is higher than we’ve seen in decades, and the WGA strike has set an example of how focusing on technology—who controls it and how it will be used—could become central to contract negotiations in other industries.
It is crucial, however, that the renewed energy of organized labor is channeled in the right way. Labor cannot hope to ask for and obtain high wages unless technology moves in a more pro-worker direction. Nor can one reasonably expect that opposing new technologies, including AI, will be a viable option for the union movement. Rather, leaders and rank-and-file workers alike must understand the potential of new tools, including artificial intelligence, and articulate a vision for how these technologies can be used in a way that helps both labor and corporations. Making workers more productive in their decision-making and problem-solving tasks is the bedrock of such an approach.
None of this will happen automatically. Any meaningful change must start with a recognition of the problem—technology is heading in an anti-labor direction. People promoting change must also recognize the feasibility and desirability of a pro-worker direction for technological change.
Neither the tech industry nor other large corporations will undertake a huge course correction without strong pressure from others in society. In addition to robust labor voices, government regulation and civil society organizations have previously played a critical role in pushing companies in the right direction. This role is even more important today when we need a fundamental redirection of AI technology.
The government should adopt a set of complementary policies encouraging a better path for technology use and development, as we detail in a recent policy memo (jointly written with our MIT colleague David Autor, through the new MIT Shaping the Future of Work Initiative that we collectively co-direct) entitled “Can We Have Pro-Worker AI?”
Lawmakers should revise the federal tax code to equalize the tax burden across labor and machines, so companies are encouraged to hire, train, and retain human workers. Regulatory agencies should also find ways to include and amplify worker voice on how AI and other technology is used in the workplace—including regulating AI-assisted personnel management and placing safeguards around workplace surveillance.
The public sector should invest in research that prioritizes human-complementary AI technology and create a federal government consultative AI center of expertise to support lawmakers and officials who need to understand this technology. We should use this federal expertise to assess whether specific AI technologies live up to their promise to augment human work before they are rolled out in any publicly funded programs such as education or health care.
There is much to be alarmed about when it comes to AI. But if AI brings a dystopian future, it will not be because this was the only path available to us. It will be because we failed to grasp how this technology could be developed and used, and how it could help workers, rather than just replacing them. Understanding this is a first step toward the right type of AI revolution.
How can data-driven technology be harnessed to give power back to workers? Listen to a case study from SSIR’s 2023 Data on Purpose conference:
Support SSIR’s coverage of cross-sector solutions to global challenges.
Help us further the reach of innovative ideas. Donate today.
Read more stories by Daron Acemoglu & Simon Johnson.