By: Rooz Aliabadi, Ph.D.
In the 2008 film “Wall-E,” humans inhabit a society that can be described as a fully automated luxury communism. In various forms, AI-powered robots perform all labor-intensive tasks. At the same time, people lead inactive lifestyles, growing overweight and engrossed in television and computer games. Taking it further, the “Culture” series by Scottish novelist Iain M. Banks fantasizes about a future where AI has evolved to superintelligent levels, surpassing current imaginings. These books have become popular among influential figures like Jeff Bezos of Amazon and Elon Musk of Tesla. In Banks’ world, scarcity is eradicated, with AI “minds” overseeing most aspects of production. Humans, freed from mundane matters, engage in artistic pursuits, explore diverse cultures across the vast universe, and abandon unapologetically self-indulgent pleasures.
The possibility of such narratives might have once seemed unlikely. Nevertheless, the remarkable advancements in generative AI, like the technology powering OpenAI’s popular chatbot, ChatGPT, have compelled many to take them more seriously. OpenAI’s founders recently stated that AI systems will surpass expert-level proficiency across various domains within the next decade, effectively engaging in productivity comparable to today’s largest corporations. Notably, last year, tech-savvy forecasters on the online prediction platform, Metaculus estimated that it would take until the early 2040s to develop an AI capable of convincingly imitating a human in a two-hour chat, possessing adequate robotic skills to assemble a car and passing other demanding cognitive tests. However, due to a year of incredible AI breakthroughs, Metaculus forecasters believe the early 2030s will reach this milestone.
The journey toward achieving a general AI surpassing human capabilities in all aspects may take longer than initially anticipated. However, the increasing plausibility of such powerful AI reminds us to consider what the future holds for humanity in its wake. Will humans, like the couch potatoes depicted in the “Wall-E,” cave to laziness and inactivity? To explore this question, let’s embark on a thought experiment guided by economic principles to offer insights into possible outcomes.
Unleashing the Potential of AI
Navigating through this thought experiment entails assuming certain optimistic premises. Firstly, we envision AI as compassionate, controllable, and dissimilar from humans. Moreover, we anticipate that human culture will remain intact despite technological advancements, allowing people to develop affection or even reverence for AI. Rather than perceiving AI as entities to cherish, we view them as powerful tools — virtual, brilliant, and cost-effective bots. We acknowledge that challenges such as energy limitations impeding the widespread adoption of AI must be overcome, although no guarantees exist. However, setting aside these uncertainties enables us to embark on this exercise.
In 2019, economists Philippe Aghion, Ben Jones, and Chad Jones constructed a model to analyze the impact of AI. Their findings suggested that explosive economic growth could be attainable if AI were leveraged to automate all production, including the research process itself, thus enabling self-improvement. This approach would allow countless AI systems to collaborate on any problem, unlocking enormous scientific possibilities. However, the economists issued an important caveat within their modeling. They emphasized that significant growth would not materialize if automation covered most but not all products or if it automated most but not all aspects of the research process. They aptly put it, “Economic growth may be constrained not by what we do well but rather by what is essential and yet hard to improve.”
Insights from Economist William Baumol shed light on this phenomenon. In a seminal paper published in 1965, Baumol and his colleague William Bowen analyzed wages in performing arts. They observed that the “output per man-hour” for a violinist performing a Schubert quartet in a traditional concert hall remained relatively steady. While technological advancements boosted productivity in numerous industries, the performing arts remained essentially unchanged. As prices rose, individuals continued to invest in these skills due to the inelastic nature of demand. Consequently, the arts sector consumed a more significant portion of GDP, exerting a drag on overall economic growth.
Baumol’s example underscores a broader principle, where the full automation capabilities of AI in certain domains create imperfect substitutes for those that remain non-automatable. Consequently, the demand for industries that cannot be fully automated becomes resistant to change. This phenomenon leads to the growth of unproductive sectors, resulting in a larger share of GDP allocated to these sectors and consequently reducing overall economic growth. Aghion, Jones, and Jones further observe that this pattern has unfolded over the past century. Technological advancements have automated significant portions of agriculture and manufacturing, leading to a decline in the relative prices of their outputs. Consequently, people have allocated more of their incomes to industries like education, healthcare, and recreation, which have yet to experience comparable productivity gains.
However, what if ultra-powerful AI also develops super-humanoid robots capable of meeting our material needs? In such a scenario, the capabilities of machine hands could replace the need for human effort. Humanity might embrace a lifestyle like that portrayed in “Wall-E,” where labor is abandoned. An essay titled “Economic Possibilities for Our Grandchildren,” by economist John Maynard Keynes in 1930, examined this idea. Keynes speculated that people would work less than 15 hours per week in a century. He envisioned that technological advancements would solve the “economic problem,” enabling individuals to devote their time to intrinsically pleasurable activities. While we have not yet attained Keynes’s envisioned 15-hour work week, increased levels of wealth have indeed contributed to a reduction in working hours, aligning with his expectations. In the rich world, the average number of hours worked per week has declined from approximately 60 in the late 20th century to below 40, reflecting the growing allure of leisure.
Despite the progress in supercharged AI, there remain particular desires that only humans can fulfill. Acknowledging that what brings inherent pleasure may also encompass work is essential. Within this context, three key areas emerge where humans may continue to play a significant role: work intertwined with play, the pure play itself, and work domains where humans possess a distinctive advantage.
The Joyful World of Play
Let’s begin by exploring the increasingly blurred border between work and play. While working hours have declined over the past century, the most significant declines occurred before the 1980s. Interestingly, it is now regarded that wealthy people work longer hours than their less privileged counterparts. Keynes’s essay offers a potential explanation for this fascinating trend. He proposed a division of human desires into two categories: absolute needs, which we feel irrespective of our fellow human beings’ circumstances, and relative needs, which we experience only when their satisfaction allows us to surpass and feel superior to others.
Keynes may have underestimated the significance of this second category of desires. It can be argued that entire academic disciplines, for instance, fall within this realm — existing seemingly detached from tangible value, with academics engaging in fierce competition to establish their intellectual superiority. From an economic perspective, work has transformed into a “consumption good” for many individuals, providing satisfaction and utility that surpasses the income it generates.
Games provide further insights into why people may only partially cease working. Millions of individuals are employed within entertainment and sports, engaging in activities that some perceive as immaterial but still compete for recognition. While it may be expected that interest in watching such games would diminish once AI surpasses human capabilities, evidence from sports where humans are already outperformed suggests otherwise. Following IBM’s DeepBlue defeating Garry Kasparov, the world grandmaster, in chess in 1997, interest in the game has increased. Similarly, other games that AI has “solved,” such as Go, an ancient Chinese board game, and competitive video games, have experienced a similar pattern. The global number of video game players has nearly doubled in the past decade, reaching 3.2 billion last year. Today, a growing class of gamers competes or stream their gameplay for a living.
AI has the potential to amplify this interest greatly. As speculated by Banks, humans could specialize in pursuits that hold deep significance in life, such as sports, games, romance, studying ancient languages, exploring societies of different cultures, tackling impossible problems, and even engaging in daredevil activities like mountain climbing without safety harnesses. Naturally, other individuals would likely be keen to observe and engage in these endeavors.
It appears unlikely that people would forfeit control of politics to robots. Instead, once AI surpasses human capabilities, people presumably pay closer attention to political matters. While specific political tasks could be delegated to AI models that generate proposals to balance individual preferences, the participation of humans in political procedures holds immense importance. Political philosophers, spanning from John Locke in the 17th century to John Rawls in the 20th century, have argued that the active engagement of citizens in political processes lends legitimacy to the outcomes in the eyes of their fellow citizens. Moreover, more negative factors would come into play as well. Humans inherently desire to influence one another, even in a world where machines fulfill everyone’s needs and desires. Statistical evidence indicates that the wealthiest 1% of Americans exhibit two to three times higher political participation rates than the general public across various measures, ranging from voting to time devoted to political activities.
Finally, let’s explore the domains where humans possess a unique advantage in delivering goods or services — a “human premium,” so to speak. This premium would ensure demand for human labor remains even in an era of highly advanced AI. One area where this advantage could be evident is transforming private information into public knowledge. As long as individuals are more inclined to share their secrets with fellow humans rather than machines, there will be a crucial role for trusted individuals who can selectively unveil this information to the world, making it accessible for AI systems to process. It is reassuring to think that investigative journalists, for instance, will continue to play a vital role.
The concept of the human premium extends to other realms as well. Humans inherently appreciate the significance of history, myths, and the deeper meaning attached to various aspects of life. This is evident in non-fungible tokens (NFTs), where the value of an item can significantly increase when its provenance is verifiable on a blockchain. Even if two images possess identical pixels, the one with a unique history is often valued at a much higher multiple.
In domains such as caregiving and therapy, humans derive immense value from the presence and time invested by others. The personal touch and emotional connection fostered in these interactions are irreplaceable. On the other hand, artificial diamonds, which possess the same molecular structure as naturally occurring ones, trade at a substantial discount — approximately 70% less, as some sources estimate. In the future, items bearing the “made by a human” tag could potentially hold exceptional desirability, recognizing the unique value that human craftsmanship and involvement bring to the table.
The Human Premium
If the human premium proves to be substantial, it has the potential to exert a drag on economic growth. By categorizing sectors of the economy based on the presence or absence of a significant human premium, we can observe the potential impact. If individuals do not substitute machine-produced goods and services for those created by fellow humans, the Baumol effect would intensify, potentially leading to a scenario where measured economic growth approaches zero. If highly advanced AI fails to supercharge growth, it would indicate that the economy has already transitioned beyond materiality, placing greater emphasis on domains such as play, politics, and areas where human interaction holds the most incredible value.
Inevitably, there will come a time when AI generates new goods and services that surpass the desire for human connection and interaction. The outcome of such a contest would reveal something profound: the extent to which humans are fundamentally social animals. It would shed light on our inherent need for social bonds and the degree to which we prioritize interpersonal relationships in our pursuit of satisfaction and fulfillment.
This Chat GPT Lesson Plan and others are available FREE to all educators at edu.readyai.org
This article was written by Rooz Aliabadi, Ph.D. (firstname.lastname@example.org). Rooz is the CEO (Chief Troublemaker) at ReadyAI.org
To learn more about ReadyAI, visit www.readyai.org or email us at email@example.com