The coming “intelligence” panacea is a certain scare for worker’s wages, and returns to labor. Concerns abound over job disruption, means of production, and the ethics of algorithms. While technology will accelerate the creation of new products and services, it will continue to shrink production timelines, supply chains, and knowledge sets. Technologies like artificial intelligence serve not only as an example of the power of capitalistic systems, but the way forward in an evolving marketplace where day-to-day life will become increasingly automated. We are slowly shifting towards a new model: one of economic altruism.
In capitalistic systems, market competition reduces the marginal cost of any good or service to its equilibrium point (marginal revenue = marginal cost). Our economy has advanced from one dominated by agriculture and manufacturing of physical goods, to one of professional services, technology (hardware and software), and information (our current position).
In each case, the commoditization of goods and services through market competition reduced the consumer costs. Capitalism as a system is a form of network architecture unlike any we have ever seen, and just like network architectures, it can expand, contract, be wasteful, complex, or efficient. The sub-optimality of capitalism is a function not of its limited outputs, but rather of the limited inputs.
The Fixed Costs of Intelligence
The “information age” of capitalism has evolved in three epochs: (1) big data, (2) information processing, and (3) intelligence (AI). One of the world’s top investors has stated that ‘the next decade will belong to companies that can turn data into useful information.’ Extending his statement, when we next organize information in a smart way and target its usage with the requesting user, we are creating “intelligence”. The resulting technologies, based on artificial intelligence, machine learning, and the like, will transform fixed costs into variable costs (which already sit at zero for technology companies due to the code repeatability). An early and rudimentary example of this is Turbotax, where software is replacing human tax experts. It’s no wonder that workflow technologies are so often associated with the rise of AI; they are often the digital backbone of intelligence and intelligent systems.
This shift from fixed cost to variable cost (both up front, and increasingly at scale) means AI will reduce the marginal cost of intelligence to zero. Extraordinarily elegant solutions that replicate human-like intelligence will be produced at a fraction of the price. Service industries worldwide will experience automation anxiety, and workers will begin to feel the same pressures that factories experienced in the 1950s. Headlines have already begun to reflect cognitive automation in fields such as law, medicine, and accounting.
The remaining fixed costs associated with the development of AI include (1) the cost of “training” your algorithms with data (collecting, processing, and transforming that data), and (2) software development time (viewing and interpreting the transformed data). While the former can be accomplished via human labor, crowdsourcing, or mechanical turks, these development costs are still largely manual. However, even these costs will become irrelevant as research drives us towards new algorithmic strategies and outputs.
Currently, the field is dominated by supervised learning algorithms, or those that require “training data” to predict outcomes, classify data, and create insights.
Unsupervised learning in turn requires little to no training data, removing additional fixed cost. Accordingly, the marginal cost of a unit of intelligence will be a product of the application and implementation of an intelligent algorithm to a noticed problem.
Capitalism’s incredible and endless march towards cost efficiency will allow individuals and companies to create constructs that mirror intelligent thought (via unsupervised learning systems) and apply them to any industry. The result? A marginal cost of zero to produce a unit of intelligence on par with human cognitive thought.
The remaining cost for “intelligence tasks” is the fixed cost investment. Innovation will shift some costs from fixed to variable (driving marginal costs for a unit of intelligence continuously downward), and investment expenditures will become nominal inputs to incredible outputs and profits.
As advanced technologies take over manufacturing and construction processes (3-D printing and drone construction), transportation (self-driving cars), food and water production (robotic gardening, closed-loop systems of food and water production), education (online educational portals), it will continue to decrease the cost it takes to “live.”
What about the Jobs?
The job disruption will be profound, as long-term efficiencies can be the result of short-term economic pain. Are jobs required for productive capability[1], or rather a contemporary construct that evolved from combinations of ancient feudalism and capitalism that serve as a stopgap until our machines and algorithms work for us? By demanding that labor be an input to our economic structures, we ignore the fact that when it is free to live in society, jobs aren’t necessary, and government welfare or expenditure will be drastically reduced. Returns to capital have increased over the past 50 years, and will continue to move in this direction as AI proliferates[2].
If the purpose of a job is composed of (1) individual self-worth and (2) obtaining the means to purchase goods and services, this future will instead empower choice (money from jobs will not be required to purchase goods and services as their cost will be near zero). Only the pursuit of individual self-worth, surely available via non-employment forms of enlightenment, will be sought.
Artificial Intelligence is thus proof that capitalism will turn the marginal cost of day-to-day living to zero dollars. Economic liberty gives an individual the ability to choose their level and type of production in a society that requires none of it while still meeting each person’s individual demands and preferences for goods and services[3]. Capitalism then, is the shepherd of liberty.
Evolving Capitalism
The conclusion, therefore, is that capitalism has worked exactly as intended. As a vehicle, it has taken the inputs costs of goods and services and driven them to zero through competition. There are two results that require review: changes to the nature of work, and how capitalism will evolve.
AI’s proliferation need not be a bellwether for a dystopian future. In a world where work is automated, the cost of those goods and services will also decrease to near zero (if inputs costs decrease, output prices decrease).
As we approach a world in which goods and services trend towards nominal or free for its citizens, we have an unparalleled opportunity to augment our economic structures. Simply put; capitalism supports altruism.
Chart 1: Comparison of Types of Political Economy
Data is the fuel that will drive it there. Three questions result. First, how do we measure? Second, how do we obtain the data? Third, how does it work?
Production of goods and services are organized by costs; raw materials, data, human labor, capital requirements. Each of these is passed through to one of the simplest yet most effective forms of information translation: price. Price contains supply chain information, shipping delays, errors, innovative production processes, intellectual property, superior employees, and more. Nonetheless, the conclusion above supports that each of these would be reduced to near zero. If the marginal cost of inputs for goods or services are zero, and we have economic competition[4] (marginal revenue = marginal cost), the conclusion is that revenues will also be zero. How do we augment capitalism to continue to progress?
The answer is fascinatingly simple. The more data we collect throughout the threads of society, the more we may input the results into our economic models. Attempts are under way to drive the cost of pollution to zero via cap and trade and other theoretical constructs.
Instead of taxing cigarettes, societies may tax quantified racial bias, gender discrimination, pollution, and other socially inefficient practices, thereby shifting these from social costs to economic costs. The construction of these incentives, however, is critically important in order to avoid negative externalities.
Once we are able to transform social costs into economic costs, they would become an input to price. In other words, data will allow us to view social costs and economic costs as one and the same.
As with any tax, this would increase the marginal cost, and thus the price of a good or service served to customers. Accordingly, capitalism would take over to drive their impact to zero. Imagine a future where discrimination is competitively eliminated, bias no longer exists, and businesses are aligned with social and political goodwill? This was previously impossible given the cost it would impose on businesses, but the evolution of AI and availability of data will allow us to drive down costs, calculate these impacts, input them into economic models, and compete them to zero.
Towards Economic Altruism
Our increasing reliance on data services (and more recently intelligence services and automation) churn out more data about users than any time in history. Using this data will allow us to augment our societal-level behaviors much in the way businesses use data to solve commercial problems, and input these costs into pricing models. By doing so, we will “solve” social issues the way capitalism has solved the prices of goods and services. The marriage between the digital and analog world is under way, and the benefits to society will be pronounced and exceedingly positive. For the first time in history, we will be able to economically incentivize morally conscious actions to drive progress as a collective humanity.
[1] The individual benefits and self-worth notwithstanding.
[2] It is a possibility that this trend will accelerate until returns capital make up nearly 100% of productive gains in our society due to the use of technology like AI.
[3] An important caveat to the outlook of a society that maximizes liberty is that AI must live in an open and free system, not be controlled by select few. That our current patent laws that prohibit the legal capture of mathematical formulae are oddly prescient in ensuring this future.
[4] Imperfections in competition models aside.
Comments