U.S. technology companies have poured an astonishing estimated $400 billion into artificial intelligence (AI) infrastructure this year alone. Projections suggest that global spending could skyrocket to a staggering $3 trillion by 2028, prompting critical questions about the potential returns on such investments and the very nature of the technology itself.
Could AI fundamentally alter the balance between capital and labor? Might it trigger unparalleled economic growth? And is AI inherently an extractive technology? These are just a few of the pressing questions that emerged during a recent discussion with Adam Tooze, an economics columnist, on our podcast, *Ones and Tooze*. Below is an edited excerpt of our conversation, which delves into the implications of AI as a “general-purpose technology” and its potential to reshape our economy and society.
Cameron Abadi: Economists often label AI as a “general-purpose technology.” What are the economic implications of this designation?
Adam Tooze: This is an excellent starting point, as it gets to the core of how economists are attempting to understand AI’s relationship with the economy and the historical context of technological change. The term “general-purpose technology” refers to innovations that initially have subtle impacts on the economy but eventually become indispensable. Examples include steam power, electrification, and semiconductors. The question regarding AI is whether it is merely a general-purpose technology or something more profound—perhaps even the culmination of technological advancement, as it applies technology to the process of thinking itself.
Most economists agree that AI fits the mold of a general-purpose technology. However, there is an open debate about whether it might represent an unprecedented leap in technological evolution. If AI is indeed this groundbreaking technology, it would radically alter economic theories about technological change.
Traditional economic growth models suggest that the combination of capital and labor leads to productivity growth, eventually leveling off as economies reach maturity. Here’s the twist: if AI acts as an innovation-enhancing tool akin to the invention of the research lab, we could witness not just temporary boosts in growth but a sustained era of accelerated innovation. This could lead to growth rates that far exceed historical norms, as we innovate more rapidly and efficiently. The crux of the debate is whether AI is just another general-purpose technology or a unique break in historical development.
CA: It seems there’s a lot of speculation about AI’s future trajectory based on its current growth. But could we already be witnessing a slowdown in AI improvements? Are we experiencing diminishing returns?
AT: This is a contentious topic within the AI community, and the debate is intense. The discussion revolves around the concept of scaling laws, which suggest that as we increase data, parameters, and computational power, AI performance improves. However, some believe that we may be approaching a ceiling. Senior figures, such as Meta’s chief AI officer, argue that even if we are not hitting a literal wall, current AI models face constraints that limit their potential. For instance, if these models are predominantly text-based, they may not be incorporating enough diverse data to achieve a qualitative leap in performance. This realization has significant implications for corporate strategy and investment, as firms explore ways to diversify their data inputs and enhance the capabilities of their AI systems.
CA: Is it fair to view AI as an inherently extractive or exploitative technology that exacerbates environmental issues and social inequalities?
AT: This argument carries weight, especially when we consider that AI is built on both natural resources and human labor. Kate Crawford’s observation that AI is neither entirely artificial nor intelligent highlights this reality. AI relies heavily on energy, water for cooling, and land for infrastructure. Moreover, while AI is often perceived as a force that displaces labor, it relies on substantial human input to function effectively. The process of anchoring AI’s capabilities in “ground truth” necessitates significant human labor to validate and contextualize the data that AI models use, often resulting in underpaid workforce exploitation.
Additionally, the methods used to gather data for training these models raise ethical concerns. Major companies have faced lawsuits over allegations of improperly using copyrighted material to develop AI systems, illustrating a troubling aspect of extraction in the field. As the industry continues to evolve, the legal battles surrounding intellectual property rights will likely shape the future of AI and highlight the need for ethical accountability in its development.
CA: Henry Farrell describes AI as a “cultural and social technology,” suggesting it has the potential to fundamentally reorder social and cultural life. Do you agree, and what aspects of society could be impacted by AI?
AT: Absolutely, I believe that’s the right perspective to take. It’s essential to move past the anthropomorphizing narratives that often accompany discussions of AI. While AI does not possess human intelligence, it is an advanced mechanism for generating symbols and can significantly influence our society. Just as the advent of writing and printed books reshaped our relationship with language, AI has the potential to alter how we communicate and interact.
AI’s ability to generate symbols and evoke emotions can lead to profound changes in our cultural experiences. For example, just as novels can evoke feelings and foster connections, AI can also shape our emotional landscape through its interactions. The capacity of AI to analyze patterns in human behavior and respond in a nuanced manner introduces new dimensions to our social fabric. The implications of this technology extend far beyond mere efficiency; they touch on the very essence of our humanity and how we relate to one another in an increasingly digital age.
As we navigate the complexities of AI’s integration into our lives, it is crucial to remain vigilant about its potential impacts on our economy, environment, and social fabric. The ongoing dialogue surrounding AI must prioritize ethical considerations and inclusivity to ensure that this transformative technology serves as a force for good in society.
