In January 2015, Congress convened the inaugural meeting of its newly-formed Sharing Economy Caucus. Led by Representatives Jerome Nadler and Eric Swalwell, a group of lawmakers, tech CEOs, and a few fortunate academics like me gathered in Washington DC to debate the future of work. Faced with an explosion of platform entrepreneurship, freelancers, and other non-traditional work arrangements, the caucus discussed how to design a more inclusive social safety net within our existing system whose funding model was predicated on a single work arrangement, regular employment. A decade later, we have taken important strides forward. Many states, including Washington and California, have improved the portability of benefits, extending parts of the social safety net to platform workers via legislative action and ballot initiatives. There is growing evidence-based understanding of the advantages and limitations of alternative funding models and guaranteed income initiatives.
Now, as we enter an era of work uncertainty catalyzed by the explosion of artificial intelligence (AI), two new challenges take center stage. As individual human capabilities are progressively embedded into generative AI systems, we must redefine the boundaries of intellectual autonomy to preserve people’s economic returns from their human capital investments, refashioning the law to answer a seemingly simple but profoundly important new question: What facets of “human capital” should be owned by humans? And as AI rapidly alters the mix of what humans and machines do, we must invest in high-quality national infrastructure for mid-career occupational transition with dignity.
A distinguishing characteristic of the generative AI technologies that have captivated our imaginations over the last two years is their ability to generate and create after being trained on what humans generated and created in the past. The machine learning breakthroughs underpinning these breathtaking innovations are thus inextricably intertwined with a sobering realization: a collection of human “works” is now no longer simply that human’s artistic or productive output, but the blueprint for an AI machine that can replicate parts of the creative process, talent, or human capital of an individual human creator. One can now easily train a generative AI system to write like your favorite author, compose music in the style of a specific musician, or generate new art that mirrors that of Picasso or Van Gogh.
And that’s just the tip of the iceberg. In the future, capturing the precise movements of the most skilled foundry worker will generate data to train their robotic digital twins. Cameras on the wrists of an expert surgeon will feed video streams as training data into machine learning systems that power skilled robotic surgeons. The email, Slack, and telephone archives of your most successful sales executives will be used to create an AI system that encapsulates critical parts of the secret sauce to their success, allowing new hires to rapidly replicate the strategies of the more experienced without actually understanding or learning them.
These new technological possibilities have profound implications for employability, for inequality, and for our humanity. As I explain in my 2016 book, “The Sharing Economy,” societies designed to decentralize capital ownership have lower economic inequality, and while platforms portended the possibility of distributed capital ownership via crowd-based capitalism, this outcome was not guaranteed. Rather, flawed platform governance choices could exacerbate rather than mitigate inequity, which is why I argued that platform policy must be designed to proactively nudge the economy towards decentralizing who holds society’s structural capital.
Correspondingly, today’s AI technologies could decentralize the availability of an array of skills and productive capabilities, empowering millions to pursue an entrepreneurial path while catalyzing the emergence of an entirely new generation of AI-enabled professionals, from educators and healthcare providers to investment advisors and data scientists. However, if a person is unable to assert some level of ownership over the blueprint of their own individual generative process, talent, or expertise, we risk a future in which intelligence and skills are commoditized and centralized excessively, leaving humans unable to enjoy the economic returns from their own human capital investments.
Enlightened societies have recognized for centuries that human dignity is closely tied to an individual’s ability to claim ownership over the “fruits of their labor.” Today, our identities are increasingly associated with our productive capabilities. As the boundaries around human capital autonomy blur or dissolve, this won’t just dampen our engagement with the economy; it could redefine our personhood. And even when imagining an especially dystopian future scenario for human participation in the economy — a Matrix-like world in which the primary value of today’s human output stems from using it to train tomorrow’s AI systems — diminished incentives for investments in human capital could lead to the loss of this essential input, leading to model collapse, a dramatic drop in the performance of AI that manifests if its training data has an inadequate fraction of human-generated content. Thus, paradoxically, the seemingly countervailing objectives of preserving human dignity and accelerating AI ascendance may be aligned, with both requiring humans to retain sufficient ownership over their human capital.
What is the solution? Addressing the challenge that generative AI poses to human intellectual autonomy is complex. Copyright law has long posited that artists should own their works rather than their creative process —- their expressions rather than their ideas — because innovation thrives when judicious imitation is permitted. However, this societal balance did not anticipate machines that could seamlessly and rapidly replicate specific individuals. A legislative redrawing of the scope of ownership over an individual’s human capital requires a careful and nuanced reconsideration of these boundaries between style and expression, with little precedent to guide the right balance. Lawmakers might alternatively consider mandating a new market that gives workers some capital ownership over systems trained exclusively on their data. A highly skilled factory worker or surgeon may then accede to the embedding of their expertise into a series of AI machines (perhaps retiring on the returns from this “capital” investment). However, the specific and relative contribution of a particular person’s training data on the output of an AI system may be hard to ascertain, and even harder to contract on.
As we grapple with the shifting boundaries of our intellectual autonomy, we must, in parallel, design and build a system that allows people to invest in, adapt, and grow their human capital to stay competitive in the rapidly evolving world of work. We have recognized for many years that advances in AI will force a significant portion of the US workforce to shift to entirely new occupations mid-career, but meanwhile, our educational funding and infrastructure investments remain squarely focused on early-career needs. Our four-year university system is indisputably the world’s best, preparing millions from around the world for the transition from high school to one’s first occupation. We need a comparable system that supports and enables displaced workers to prepare for and transition to their future occupations.
To succeed, a holistic ethos must guide the design of this new mid-career transition education system. The global leadership and continued success of America’s college system is predicated on the realization that transition education is not simply about “skilling.” Skill-based classroom training is just one part of the university experience, and the value of a college degree comes at least as much from what marketable work skills it is bundled with — critical thinking, learning how to learn, the ability to explore potential career paths through student clubs, the formation of one’s future professional network, mentoring from peers and advisors, placement services, the credentials that accompany the degree, and the celebratory rite of passage that sees college as an accomplishment and an investment rather than a burden on society. We need a multifaceted educational infrastructure of comparable quality for mid-career occupational transition that creates new, focused, shorter-duration packages of education, bundling reskilling and upskilling with talent discovery, mentoring, networking, confidence building, branded credentialing, and job placement. This will enable our workforce to transition to the future of work with dignity. And of course, we must also address the foundational issue of human intellectual autonomy, so that these new investments in human capital pay off for the individuals who make them.
In 2009, social media was exploding and the personalization of content was starting to seed the filter bubble. However, it took many years for the polarizing dangers of hyper-personalization and misinformation to gain mainstream policy attention. By the time the issue was on the legislative front burner a few years later, the associated business models were too mature and platform users too entrenched for a bottom-up and forward-looking policy solution, forcing us to settle for partial policy fixes that continue to be reactive rather than proactive.
We are at a similar moment today. The AI-fueled workforce transition has just begun. Generative AI is still in its infancy, and not yet deeply embedded into the economy, our workplace, or our daily lives. It is crucial to look ahead and act now. We can preempt devolving into a de facto regime in which humans cede control over their intellectual autonomy, and widespread inequality of access to economic opportunity further exacerbates disenchantment and polarization. But if we fail to proactively design the world we aspire to live in, other forces will design it for us, and we may not like the world we end up in.
About the Author
Arun Sundararajan
Harold Price Professor of Entrepreneurship and Director,
Fubon Center for Technology Business and Innovation,
NYU Stern School of Business
Arun Sundararajan is the Harold Price Professor of Entrepreneurship and Technology at New York University’s Stern School of Business, where he also serves as Director of the Fubon Center for Technology, Business and Innovation. An internationally recognized expert on artificial intelligence governance and the future of work, his best-selling and award-winning book, “The Sharing Economy,” published by the MIT Press, has been translated into Japanese, Korean, Mandarin Chinese, Portuguese and Vietnamese. Find him at https://digitalarun.io/ and https://linkedin.com/in/digitalarun
About this Series
This is part of a series called “Back to the ‘Future of Work’: Revisiting the Past and Shaping the Future,” curated by the Aspen Institute Future of Work Initiative. For this series, we gather insights from labor, business, academia, philanthropy, and think tanks to take stock of the past decade and attempt to divine what the next one has in store. As the future is yet unwritten, let’s figure out what it takes to build a better future of work.
