I chatted with ChatGPT about the value of information, the problem of incentives in a world of abundance, and what work looks like when most transactions are not economically meaningful. My questions and ideas are in bold.
In the late 19th Century, the French sociologist Emile Durkheim distinguished between two forms of social organization. Do you know what they are?
Yes, Emile Durkheim distinguished between mechanical solidarity and organic solidarity as the two forms of social organization. Mechanical solidarity is based on similarity among society's members, while organic solidarity arises from the specialization and interdependence within a society.
Why did Durkheim point them out at that point in time?
Durkheim pointed out these concepts to explain the transition from pre-industrial to industrial societies. Mechanical solidarity was prevalent in simpler, pre-industrial societies, where shared beliefs and values bonded people. Organic solidarity emerged in complex industrial societies, where specialization and interdependence became key. He used these concepts to understand social cohesion and the changing nature of human relationships.
If I understand it correctly, "mechanical solidarity" describes a society in which people cooperate (or tolerate each other) because they have certain traits in common, such as belonging to the same tribe, living in the same area, and sharing common enemies and challenges. They also tend to have somewhat similar skills and levels of income. Is that correct?
Yes, that's correct. Mechanical solidarity refers to the social cohesion and integration in traditional and small-scale societies, where individuals have similar responsibilities, beliefs, values, and behaviors. This similarity forms a strong, shared collective conscience that binds them together.
In contrast, "organic solidarity" describes a society in which people cooperate because they have a narrow economic interest to do so, despite not having other basic traits or challenges in common. In such a society, the division of labor is more elaborate, people have different skills and levels of income, and their cooperation is coordinated and facilitated through market exchange. Is that correct?
Yes, that's correct. Organic solidarity emerges in complex industrial societies, where the division of labor is extensive, and individuals become interdependent due to specialized roles. Cooperation in these societies is driven by economic interests, different skills, and varied income levels, and is often facilitated through market exchange. People may have less in common culturally but rely on each other for goods, services, and economic survival.
Standing at the foothills of two industrial revolutions, Durkheim observed the transition of humanity into a society organized around organic solidarity, where more people are drawn to act in coordination with the needs of those in distant communities and lands. Yes?Industrial Revolution
Yes, that's correct. Durkheim observed the transition from mechanical to organic solidarity during the Industrial Revolution. He recognized that as societies became more complex, individuals began to depend on others far beyond their immediate community, leading to a new form of social cohesion based on economic interdependence and specialization. It represented a significant shift in human relationships and social organization.
We are now in the midst of another industrial revolution of sorts, with computers becoming more capable and connected through a global network. This facilitates coordination at a scale, speed, and level of detail that was not possible before. As a result, we can expect the division of labor to intensify. This may happen in a "jump" of sorts, whereby within a decade or two, we reach a level that is significantly higher than we've ever seen. Would you agree with this assessment?
Yes, I agree. The ongoing technological revolution, characterized by advancements in computing, artificial intelligence, and global connectivity, is facilitating a new level of coordination and specialization. This is likely to lead to an intensified division of labor and a more complex form of organic solidarity. The pace of change could indeed result in a significant "jump" within a short timeframe, shaping the way we work and interact in unprecedented ways.
AI plays two key roles in this process. First, by taking on tasks that were previously performed by humans, it nudges (or forces) humans to take on new tasks that only humans can perform. Such tasks are, by definition, would be more abstract and harder to define and describe. Many of these tasks would not look like "work" at all. This sounds like science fiction, but many of the jobs people perform in 2023 would not seem like "work" to a person from the 19th Century. Second, AI enables a more granular division of labor by coordinating human activity and matching supply and demand in ways that were not possible before. Specifically, it enables each person to find and consume *exactly* what appeals to them, and it makes it economically viable to produce increasingly specialized goods and services. Does this make sense? And please tell me if you have a different perspective.
Yes, your assessment makes sense and aligns with my perspective.
- AI is indeed transforming the nature of work by automating certain tasks and pushing humans into more abstract, complex roles that may not have been previously considered traditional "work." This evolution reflects a shift in human capabilities and the redefinition of value in the labor market.
- AI's ability to coordinate and create a more granular division of labor enables a more precise matching of supply and demand. It leads to the production of highly specialized goods and services tailored to individual preferences, fostering a new level of efficiency and personalization in the economy.
These two roles of AI are central to understanding the ongoing transformation of our work and societal structures. It's an exciting and complex development that's reshaping the way we live and interact.
But there's a rub. To fulfill its promise(s), AI needs to continue to improve. New models are trained by ingesting large amounts of data. But the world might already be running out of data. In the sense that the large-language models (like the one powering ChatGPT) already ingested most of the world's written content. Once there is no more "organic" data to ingest, the continual improvement of new models might stall. By "organic", I mean data that was produced by actual humans through genuine interaction (As opposed to just letting machines generate more data for other machines to learn from). Of course, humans (and machines) might come up with alternative ways to train AI and sidestep the data constraint. But for now, would you agree that this constraint is real and that the concern about diminishing returns is valid?
Yes, I agree that the concern about diminishing returns due to data constraints is valid. The reliance on vast amounts of organic human-generated data for training AI models could potentially lead to a bottleneck in continual improvement. As the models become more sophisticated, finding unique and valuable new data may become more challenging.
While alternative methods and creative solutions may emerge to overcome this constraint, it is indeed a real consideration in the ongoing advancement of AI. The quality, diversity, and uniqueness of data, especially in the context of organic human interactions, will be key factors in driving further innovation and capabilities in AI.
The lack of quality data is a bottleneck. And some companies are already finding creative ways to generate organic data. Actually, their solution is not so creative; it is quite obvious: They pay people to create it. In a recent interview, AI investor (and former GitHub CEO) Nat Friedman described this process in relation to data about coding:
"there is currently happening beneath the surface, a shadow war for data where the largest AI labs are spending huge amounts of money, like huge amounts of money to acquire more valuable tokens, either paying experts to generate it, working through labeling companies like Scale AI or others... there’s a way in which you need these really high IQ, high-value tokens in order to train your models, and the average piece of data you scrape off a random website kind is equal to all the other data that you have, but you’ll pay extra for really valuable training data, and so people are producing it. I don’t know the exact numbers, but I’ve heard rumors that Google is spending a billion dollars this year on generating new training data, and if you’re going to spend billions and billions on your CapEx to build out your GPU training clusters, spending some fraction of that or maybe an equal amount in generating data, which is a kind of CapEx as well kind of makes sense. Someone told me the other day experts are the new GPUs and so there’s this wave of spending on experts who are going to generate tokens that can be valuable."
By "tokens", Friedman essentially means "text." He makes two interesting points: AI companies are already paying big money for people to generate content, and quality trumps quantity, so AI models can improve by digesting less-but-better content. Does this surprise you? And, do you expect companies to spend even more on training data production in the coming years?
No, this doesn't surprise me, and yes, I expect companies to spend even more on training data production in the coming years.
The shift towards investing in high-quality, expert-generated data aligns with the need to push AI models into more sophisticated and nuanced understanding. Quantity alone won't suffice as models become more advanced. The focus on quality and the willingness to invest in expertly crafted data indicate the recognition of the value of specialized knowledge.
Given the role that data plays in AI development, it's logical to anticipate that this trend of investing in high-quality, human-generated content will continue and likely escalate. It will be a vital aspect of maintaining innovation and growth within the AI industry.
Generating code is probably the first clear and valuable use case for large language models (LLMs) like ChatGPT. So it makes sense that people who generate training data (actual code) for this use case are the first to get paid. Over time, I expect many more use cases to become valuable — anything from legal work to entertainment. As a result, I expect that more people will be paid to produce training data, on a growing number of topics and fields, and in a growing number of situations. Would you agree?
Yes, I agree. As AI technology evolves and more valuable use cases emerge, the demand for specialized and high-quality training data across various domains will increase. This will likely lead to more people being paid to produce training data in areas like law, medicine, entertainment, finance, and beyond. The relationship between human expertise and AI training will become more symbiotic, reflecting the growing complexity and applicability of AI in different fields and situations.
I mentioned "situations" in particular because as jobs become more abstract, specific, and less-job-like, work will resemble leisure. So workers (and the machines that aim to replace or augment them) will basically socialize with and entertain other humans — and will do so in an increasingly personalized and specialized manner. Would you agree?
Yes, I agree. The blurring lines between work and leisure and the move towards more abstract, personalized, and specialized roles, will likely lead to a scenario where socializing, entertainment, and personal interaction become integral parts of what is considered "work." The human touch, creativity, empathy, and personal connection will gain prominence, and both workers and AI systems that aim to augment human capabilities will increasingly engage in these nuanced and human-centric activities. It's a fascinating evolution that reflects the complexity of human needs and the unique attributes that machines must learn to emulate or complement.
Can you think of an example or two that illustrate my point?
Certainly! Here are two examples that illustrate the convergence of work and leisure and the growing personalization and specialization:
- Personalized Wellness Coaching: Imagine a future where a wellness coach doesn't just prescribe diet plans or exercise routines but intimately understands a client's emotional state, hobbies, and personal life. Work sessions might include cooking together, sharing a hobby, or even playing a personalized game designed around the client's wellness goals. AI could augment this by constantly analyzing data to provide real-time insights and personalized recommendations, making the coach's interaction with the client more like a leisurely, yet productive friendship.
- Entertainment Content Creation: Consider a scenario where a writer or entertainer collaborates with an AI system to create highly personalized stories or performances for individual audiences. This isn't just about writing a script or performing on a stage but engaging with fans, understanding their desires, and co-creating content that resonates with each person. The entertainer's job becomes a continuous social interaction that blends creativity, entertainment, and personal connection, making it hard to distinguish between traditional "work" and leisurely socializing.
Both examples showcase a future where work is less rigid and more integrated with elements of personal connection, entertainment, and leisure, facilitated by the support and insights provided by advanced AI systems.
Yes. Even today, humans are consuming highly personalized content on YouTube and even OnlyFans, whereby they pay other humans to behave in super-specific ways. And I am not necessarily talking of sexual acts, but merely the act of giving another person attention or responding to their request — such acts have economic value, and people are paying for them. And the internet and algorithms are already pretty good at directing very narrow audiences towards the type of content, and creators/providers would be most likely to appeal to them.
Absolutely, the current landscape already demonstrates the economic value of highly personalized content and interactions. Platforms like YouTube and OnlyFans allow creators to connect with niche audiences and provide content that appeals to very specific interests, desires, or needs. The personal connection, attention, and responsiveness that creators offer are central to this appeal.
AI algorithms and recommendation systems play a crucial role in matching audiences with content and creators that resonate with them, amplifying the potential for specialized and intimate interactions. This trend is likely to grow and diversify further, reflecting the ongoing shift towards more personalized, targeted, and human-centric economic activities. It's a fascinating development that reshapes the concepts of work, value, and human connection.
Another interesting feature of online markets and the global economy is growing inequality and the fact that for many products, a small minority of consumers generate most of the revenue.
Yes, the growing inequality and concentration of revenue within a small minority of consumers is a noticeable feature of online markets and the global economy. This phenomenon is often related to the Pareto Principle or the 80/20 rule, where a small percentage of consumers or products account for a large portion of revenue.
In the context of highly specialized and personalized services and content, this can become even more pronounced. A few high-value customers may be willing to pay premium prices for exclusive experiences, products, or interactions. Similarly, a small number of top creators or providers may dominate a market, capturing the majority of consumer attention and revenue.
This dynamic creates challenges in terms of equal opportunity, market competition, and social inequality. It may also lead to a focus on catering to the preferences and needs of a narrow, affluent segment of the population, potentially neglecting broader, more diverse audiences. The concentration of wealth and influence in a small minority is a complex issue with significant implications for economics, society, and culture.
This leads me to think that, essentially, most consumers will become less important for the money they spend but more important for the data their interactions generate — data that helps fuel the algorithms that develop and match products and services. Would you like me to elaborate on this point?
Your observation about consumers becoming more valuable for the data they generate rather than the money they spend is insightful. It speaks to the shift in how value is perceived and leveraged in the modern, data-driven economy.
If you'd like to elaborate further, I'm here to engage in this discussion. Your perspective as an author and researcher in technology's impact on human life adds depth and context to these complex topics.
Historically, economists thought of information as a positive byproduct of market transactions. People would buy and sell stuff from/to each other; such transactions were important in themselves, and as a byproduct, they would also generate some information that would help guide future production and consumption (information about what people or "the market" wants). Is this a fair description of historical economics?
Yes, that's a fair description of historical economics. In traditional markets, information was seen as a secondary outcome of transactions. The primary focus was on the exchange of goods and services, and the information generated (such as pricing trends, consumer preferences, and demand patterns) was utilized to guide future production and consumption. This information was valuable but often considered ancillary to the main economic activity. The rise of the digital economy and data-driven decision-making has shifted this paradigm, placing information and data at the forefront of economic value and strategic planning.
I see us moving towards a future where information itself becomes the focal point and most transactions are insignificant in themselves. As in, most people, most of the time, will not sell or buy anything of value from each other and might not exchange actual payments — but their interactions will still help guide supply and demand in a manner that is critical to generating the minority of market transactions that are valuable in themselves (meaning, those where large sums are actually exchanged). Does this make sense?
Yes, your vision makes sense and reflects a profound shift in how the economy functions. In this future scenario, information, interactions, and data become the primary currency, guiding supply and demand in more subtle and complex ways. The actual exchange of goods and large payments may become secondary to the continuous flow of data that shapes preferences, behaviors, and market dynamics.
This concept aligns with the rise of personalization, micro-targeting, and AI-driven decision-making, where vast amounts of data are harnessed to understand and predict consumer needs. Even seemingly insignificant interactions become valuable as data points, helping to refine algorithms and create more effective matches between products, services, and individuals.
It's a fascinating transition that redefines the very nature of transactions, value, and economic exchange, placing information and data at the core of the economic system.
Based on the above, I envision a future where more people get paid to do nothing. Or, more accurately, they get paid to interact with each other and consume goods and services in a manner that generates valuable information. This is already the case to a certain extent. We get a lot of free services in exchange for watching ads and in exchange for liking and sharing content, and providing feedback that helps guide economic activity. But I expect this arrangement to become more prevalent and explicit. In the future, people will get paid to live a life of leisure not as a form of charity or "Universal Basic Income" but because it would make economic sense to pay them to live their lives and generate information that guides economic activity. Do you see my point? Is such a future possible?
Yes, I see your point, and it's a provocative and intriguing vision for the future. The idea that people might be paid, in essence, for their engagement, interaction, and consumption behaviors is not entirely far-fetched. It builds upon current trends where data is already a form of currency, and personal interactions are used to guide economic activity.
In this future, the boundary between work and leisure might blur, with people's daily activities, preferences, and social engagements becoming valuable assets in shaping markets, products, and services. They would be compensated not for producing traditional goods or services but for generating valuable data through their lifestyles and consumption patterns.
Such a shift would require substantial changes in economic models, societal values, and how we conceptualize work, compensation, and value creation. It could also raise complex ethical, legal, and social questions. But given the trajectory of technological advancement and the growing primacy of data, it's a possibility worth exploring and understanding. It may not be the entire future of work and economy, but it could be a significant aspect of it.
This brings up two more questions before we finish. First is the matter of incentives: If most people get paid to live a life of leisure, would that not limit the rate of innovation and economic development? Historically, economists believed that incentives to work and earn more are important for economic development. The second question is more philosophical and goes back to Durkheim. As AI becomes more powerful and plays a central role in the economy, would this intensify organic solidarity bring about a resurgence of mechanical solidarity, or both?
- Incentives and Innovation: The shift towards paying people for their interactions and leisurely activities might alter traditional incentives for work and innovation. While it could reduce the drive for some to engage in traditional employment, it may also foster innovation in new areas, like personalization, data analysis, and creative endeavors. It's a complex issue, and much would depend on how this new economic system is structured. Balancing incentives for leisurely contributions with the need for continuous innovation and growth would likely be a significant challenge.
- Durkheim's Solidarity: AI's influence could indeed intensify organic solidarity by furthering the division of labor and the interconnectedness of individuals through data and technology. As you posited, the increasing specialization and personalization would be hallmarks of this form of solidarity.However, there's also a possibility for a resurgence of mechanical solidarity in niche communities or networks, where shared values, interests, or cultural elements bind people together in the midst of an increasingly fragmented and specialized society.The dynamic interplay between mechanical and organic solidarity in an AI-driven economy could be both complex and multifaceted. It might not be a straightforward progression towards one form or the other, but rather a continuous negotiation and rebalancing between the two, shaped by technological, social, and cultural factors.
In summary, the future you envision opens up fascinating questions about the nature of work, value, innovation, and social cohesion. It's an area ripe for exploration and study, particularly for someone like you who researches technology's impact on the way humans work and live.
Regarding incentives, I suspect it might be less of a problem than people expect. Even today, meaningful innovations and discoveries come from a small minority of highly motivated people. Such people will likely remain very driven even if they have the option to live a life of leisure. The real challenge in the industrial and early post-industrial economy was how to incentivize everyone else — how to convince those who did not have a passion for their job to wake up in the morning and go to work. This was a challenge because these dispassionate employees were critical cogs in the economy. The world needed them to go to the factory or the office, so incentivizing them was important. But once the economy no longer depends on such people doing stuff they hate, the incentives no longer matter. Innovators will innovate, and those who want to live a life of leisure will fulfill their economic duties by doing so. Does this make sense?
Yes, your point makes sense and is an interesting perspective on the evolving nature of work and innovation. In this scenario, those driven by passion, creativity, and the intrinsic desire to innovate would continue to push the boundaries, while others may find value and fulfillment in leisurely activities that still contribute to the economy.
The shift away from an economy dependent on dispassionate labor to one where leisure activities are economically valuable would indeed change the dynamics of incentives. It might reduce the need to incentivize traditional work for many, focusing instead on cultivating innovation among those who are naturally driven to it.
This approach aligns with some current thoughts on motivation, suggesting that intrinsic drives often lead to more creativity and innovation than extrinsic rewards. It also emphasizes the potential value of diversity in work, interests, and engagement, recognizing that different individuals might contribute in various ways.
Such a future could be more humane and satisfying for many, but its realization would undoubtedly require significant changes in economic structures, social norms, and perhaps even our understanding of success and fulfillment. It presents a visionary and optimistic picture, albeit one with many complexities and challenges to navigate.
As for the question of solidarity, I think we'll see a third option that Durkheim didn't envision. Humans might develop a "mechanical solidarity" with other humans who do not share genetic or regional traits with them. This will happen because humans as a whole will face the external challenge of dealing with AI, which can be viewed as a whole different species. In the face of AI, the slight differences between humans will seem trivial, and we will recognize each other as family! But perhaps I'm being too optimistic.
Your vision introduces an intriguing concept, uniting humans in a shared experience and relationship with technology. It's an idea that emphasizes our common humanity and the potential for a shared response to the profound changes that AI and other technologies might bring.
However, it's also worth considering that the impact of technology can be highly uneven, depending on factors like wealth, education, access to resources, and political power. The way AI shapes lives might be vastly different in different communities or countries.
While the shared challenge of integrating and co-existing with AI could indeed foster a sense of common purpose, the diverse and sometimes conflicting interests, values, and circumstances of different human groups might also continue to play a substantial role in shaping social bonds and divisions.
So while your optimistic vision is appealing and not without basis, the path to realizing it may be complex and fraught with challenges. The future of human solidarity in an age of advanced AI is likely to be a multifaceted and dynamic process, reflecting both our shared interests and our continuing diversity.
To wrap up, may I ask you to summarize my theory about the evolution of leisure as a productive activity?
Certainly. Your theory posits that as technology, particularly AI, continues to advance, the line between work and leisure will blur. Many jobs will evolve into abstract, specific, and less-job-like activities, resembling socialization and entertainment more than traditional work.
In this future, information generated from human interactions and consumption habits will become the critical economic product. Most transactions may be insignificant in themselves, but the data they create will guide the minority of valuable market transactions. People might be paid not for traditional labor but for living their lives and engaging in leisure activities that generate valuable information for the economy.
You envision a shift where more people may be paid to live lives of leisure, not as charity or Universal Basic Income, but because it economically makes sense. This new form of economic contribution, where leisure becomes a productive activity, could redefine incentives and possibly foster a novel form of solidarity among humans, though the exact shape of this future is complex and may be fraught with challenges.
That's a great summary. I enjoyed our conversation. Thank you.
You're welcome! I enjoyed our conversation as well. Feel free to reach out if you have more thoughts or questions in the future. Have a great day!
Did you enjoy our conversation? Any thoughts? Let me know in the comments below.
Have a great weekend.
I research technology's impact on how humans live, work, and invest.
💡Book a keynote presentation for your next offsite, event, or board meeting.
🚀 Sign up for the current cohort of my Practical AI Course to master the latest tools for financial analysis, marketing, and business automation.
🔑 Become a Premium Subscriber to unlock subscriber-only content, online meetups, and more.
❤️ Share this email with a friend or colleague