How is the economic value going to be created via AI?
To understand how AI will change value creation, we need to look at how humans have been doing it so far.
A Quick Timeline of Economic Shifts
Hunting and Foraging: Early humans relied on hunting animals and foraging for food as their main source of sustenance. This was the initial form of economic activity, based on survival and resource gathering.
Agriculture and Farming: Around 10,000 years ago, humans began domesticating plants and animals, leading to settled agricultural societies. Farming increased food production, allowing for population growth and more complex social structures.
Industrial Revolution (Steam Age)1: In the late 18th century, the invention of the steam engine triggered the Industrial Revolution. Machines began to replace manual labor, driving mass production, urbanization, and significant economic growth.
Knowledge Economy: By the late 20th century, economic value began shifting towards knowledge and information. The rise of computers and the internet transformed how people worked, with a focus on services, intellectual property, and technology-driven innovation.
What is the economic role of computers over the past 80 years?
World War II: The early use of computers can be traced back to World War II, where they were employed to perform complex trajectory calculations for artillery and cryptographic purposes. These early applications demonstrated the immense potential of computers to solve mathematical problems more accurately and quickly than humans could, paving the way for broader applications in the post-war era.
1960s - Mainframe Era: In the 1960s, mainframe computers began to be used in corporate settings. For example, The Sabre system, developed for American Airlines in partnership with IBM. Sabre revolutionized the airline industry by automating flight reservations, drastically improving efficiency and transforming customer service2. These mainframes provided value by enabling massive data processing capabilities that humans simply couldn't handle manually.
1980s - Replacing Specific Roles: Over time, computers began to replace specific human roles. In the 1980s, the role of the flight engineer3 in a cockpit, responsible for managing various in-flight mechanical systems, was gradually phased out as automation advanced.
1990s - Administrative Automation: Similarly, in the 1990s, the administrative assistant's workload was significantly reduced with the introduction of Microsoft Office. Tools like Word, Excel, and email replaced the need for paper-based communication, manual spreadsheets, and other labor-intensive tasks. The impact was clear: computers shifted from performing niche calculations to automating entire job functions, freeing humans to focus on higher-level activities4.
What is the current value of the human in value creation?
In previous eras humans drove value via physical labor and/or repetitive tasks. In the knowledge economy, the primary output of humans is decisions. Today's value lies in making strategic, data-driven choices. For example, managers use data insights to decide on market expansion, engineers prioritize which features to develop based on user feedback, and financial analysts decide on investment opportunities by interpreting complex datasets.
This historical role—handling repetitive or precise tasks—is the key to how AI will drive value in corporations.
How should we segment this current decision making role?
In corporations today, human roles can be segmented into two primary categories: majority resource allocation and front-line task-specific roles.
Majority Resource Allocation: These roles are focused on strategic decision-making, determining where resources—such as time, capital, and talent—should be allocated to maximize value. Examples include executives setting long-term corporate strategies, managers deciding on budget allocations, and product leaders prioritizing features based on market needs. The core output is to navigate uncertainty, align organizational goals, and make impactful decisions.
Front-Line Task-Specific Roles: These roles focus on execution and specialization within specific tasks that are critical to daily operations. They include activities like doctors, customer service, project management, and direct sales, where individual contributors engage in well-defined tasks to achieve organizational objectives. While many front-line roles are being enhanced or automated by AI, human empathy, creativity, and problem-solving at a granular level still add significant value.
Why should we believe humans are inefficient at decision making?
Communication Overhead5 - Humans communicate to each other by converting thoughts to analogue audio or text. This leads to immense time loss as well as room for misunderstanding.
Recency Bias - In Principles by Ray Dalio there is significant mention of him realizing that an event hadn’t occurred in his lifetime, however, was part of a larger pattern. Similarly, there are countless business case studies of the executive facing a decision for the first time (large merger?) and lacking the experience and objective data to contextualize the decision. A machine with a near infinite database does not suffer from this issue.
Sample Bias - Human executives often make faulty pattern matches. They do this by drawing on samples that are not representative6 of the “population” of the type of decision they face.
note7
How will computers drive value from here?
To understand value creation, let’s briefly skip from the human to the “business as an entity” lens. A business has two external constituents: customers and owners. There are two ways to drive value to the customer: decreased cost or increased benefits. There are two ways to drive value to the investors: decreased cost or increased revenue.
Types of Value: Revenue Growth and Cost Savings (Investor Lens)
AI will create value in two key ways: increasing revenue and cutting costs.
Cost Take Out - What do we have to believe in order to think there is ample cost to be taken out of human decision making?
✅ Large Spend on human decision making - General and Administrative (G&A) expenses as a percentage of revenue is roughly 12% to 15% across publicly traded companies. The Upper Bound (Mean + 1 Standard Deviation) is about 19% to 25%.
✅ High probability that computers will be able to perform these decisions8
Revenue Growth - What do we have to believe there is ample revenue upside due to humans being replaced by machine driven decisions?
✅ Price will drop leading to elasticity of demand - This has been the constant from the printing press impact on the book sales, assembly line on Model T, and cell phone electronics.
✅ Product Quality will increase driving increase value chain capture - We have seen through advances in material sciences, computer design, etc. that better products are able to capture adjacent “jobs.”9
✅ Deeper Niche Penetration: As we have seen via Amazon, dropping the cost of serving various niches can increase demand for the company that can accomplish this. We also know over time that in certain cases specialized SKUs can drive further use cases10.
Notes
View this as a fast over-view…I will dive deeper into several of these items overtime as I finish building out the high level view.
Technology not only creates dramatic upswings in GDP/Capita, it also creates opportunities for countries to bypass others.
The dramatic uptick in global RGDP per capita around 1800 was driven by the Industrial Revolution (late 18th century to early 19th century), marked by the introduction of the steam engine (1760s), which revolutionized mechanized production, and the rise of centralized factory systems that created economies of scale. The use of coal as a primary energy source increased productivity, while the transportation revolution (early 1800s), including railroads, cut costs and expanded markets. The Agricultural Revolution (mid-18th century) provided food surpluses, supporting urbanization. These changes, combined with capital accumulation from colonial ventures and supportive institutional frameworks, led to sustained economic growth.
The Sabre system revolutionized the airline industry by automating flight reservations, which resulted in a 66% reduction in booking time and allowed American Airlines to handle 17,000 reservations per day compared to 5,000 before its implementation. This automation reduced manual errors and improved response times, leading to a threefold increase in efficiency and a significant enhancement in customer service capabilities.
The flight engineer historically managed the complex systems of large aircraft, such as fuel, engine settings, and electrical systems, ensuring optimal performance and safety. With the advent of Full Authority Digital Engine Control (FADEC), many of these tasks were automated. FADEC, approved by the FAA in 1988, integrates engine control into a single system that automatically adjusts engine performance based on real-time data, eliminating the need for manual oversight by a flight engineer. It monitors and adjusts parameters like fuel flow, power, and engine temperature, optimizing performance and improving safety by reducing human error.
In the 1990s, the introduction of Microsoft Office significantly reduced the workload of administrative assistants. Tasks that previously took hours, such as creating documents, managing spreadsheets, or scheduling meetings, were now streamlined using tools like Word, Excel, and Outlook.
For instance, studies showed that productivity increased by approximately 25% as administrative tasks were automated, allowing assistants to handle more complex work rather than repetitive, manual tasks (Microsoft, 1996; Gartner, 1999). This shift not only enhanced efficiency but also redefined the administrative role to focus on higher-value activities such as coordination and strategic support.
Imagine our example with the flight engineer. In the past the flight engineer could make observations / communicate with the pilots several times a minute. In contrast the FADEC can do this thousands of times a minute with little risk of transcription error.
Yahoo's Rejection of Google Acquisition (2002): Yahoo executives underestimated the value of Google's search technology, assuming that their existing search and portal model was good enough. Their evaluation was based on a sample of their existing user data and the popularity of web portals at the time, leading them to dismiss the transformative potential of a search-first approach that Google would ultimately dominate.
General Motors Ignoring Japanese Automakers (1970s-1980s): General Motors ignored the early success of Japanese automakers like Toyota and Honda, who focused on small, fuel-efficient cars. GM’s decision-making was biased towards historical data showing that Americans preferred larger vehicles. They didn't realize that rising oil prices and shifting consumer preferences demanded a different approach, leading to significant market share loss to Japanese competitors.
Clearly, this list is not exhaustive.
I’m going to save this for a future article. I have thought about this from several lenses and this is an article onto itself due to the complexity.
An example of a product handling multiple “jobs” is the iPhone. Previously one would need a separate camera, phone, calculator, laptop, pager, etc. to complete all of the jobs this device performs.
Another historical example of costs dropping to allow niche penetration: Henry Ford revolutionized the automobile industry by introducing the Ford Model T in 1908, which was the first mass-produced passenger vehicle, making cars affordable to the general public.
The Model T reached peak production in 1923, with around 2.1 million units produced that year. Later, in 1917, Ford also released the Model TT, the first truck from Ford, which adapted the passenger car design for commercial use. The Model TT achieved peak production volume in 1926, with approximately 201,000 units produced, helping to establish the truck market in the United States.