From Horses to Hallucinations
What 130,000 Horses in NYC Can Teach Us About the Future of Ai.
Cook’s PlayBooks is proud to partner with Brex, the modern finance platform built for high-performing CFOs. Brex unifies global cards, expenses, travel, procurement, banking, and bill pay into one AI-powered solution, giving finance teams the visibility, control, and speed to scale. Get started today and see how your peers are getting more leverage from finance with Brex.
Bilal Zuberi (bz) who used to sit on one of my boards and is a great VC and thinker wrote an extremely insightful post last week on LinkedIn. I immediately commented. Another commenter in the thread introduced a new stat I hadn’t heard before - the # of horses in NYC in 1900. Yes, it’s all relevant.
This stat immediately triggered Jevons Paradox for me. It was yet another example of how historical technology improvements and corresponding cost efficiencies paradoxically predict increased future demand and overall prosperity.
Sidebar on Jevons:
In 1865, William Stanley Jevons first described a paradox. He maintained that more efficient steam engines (powered by coal) would not decrease the use of coal in British factories (which used steam engines) but would actually increase the demand for coal.
As coal became cheaper (due to the sig. increased efficiency of the steam engines), demand for things powered by coal would grow, leading to the long run increase (not decrease) of the overall demand for coal.
I’ve certainly seen this paradox play out in my lifetime watching the cost of early computers, hard drives, memory, etc, become exponentially more efficient (Moore’s law) which in turn exponentially decreased the cost of computers and their components.
By the mid-late 90’s, many pundits were predicting a “race to the bottom” in terms of pricing for companies like Seagate and Western Digital (hard drive manufacturers). But Jevons was right, the lower cost of “everything compute” exponentially increased worldwide demand leading to the birth of all new technology industries, companies, and jobs. As hard drive storage costs and now cloud costs have plummeted, companies are now storing exponentially more data (photos, videos, data) which in turn requires more data centers to house all the data. More data in the cloud enabled all new faster and easier software products. Faster and more efficient led to increased usage and consumption… not less.
This post is about my optimism for our future Ai world and what we are about to witness over our next 25 years as Jevons meets Ai. Now you know why I titled this week’s post. Let’s jump into it.
“From Horses to Hallucinations: What 130,000 Horses in NYC Can Teach Us About The Future of AI”
In 1900, New York City relied on 130,000 working horses to move its people and power its horse drawn carriage economy. Yes, animals, not vehicles, hauled carriages and wagons through the streets each with a human driver, each requiring daily feeding, stabling, grooming, and, yes, scooping.
P.S. I used to have to muck my own stalls as a kid growing up in Ohio caring for horses… I know this job.
Today? There are reports of fewer than 100 working horses left in all of NYC. We must have lost hundreds of thousands of jobs when the automobile and other vehicles took over from horses, right? Wrong. We created a few million more jobs!
Horses were replaced by the steam engine, then combustion engines, then electric motors, and increasingly now software code that runs most of NYC.
This is Jevons Paradox at work. These are the 2nd and 3rd order effects of technological change… a paradox that most people don’t recognize until well past the normalization each technological change brings into their personal lives.
When automobiles first emerged in the early 1900s, many assumed they would reduce the congestion of horses and displace a lot of horse drivers and pooper scoopers. Quite the opposite. With faster, more convenient transport, more people traveled. The same goods moved farther and faster than before.
Suburbs were born (you’re welcome Long Island and New Jersey!) as people had the means and the method to now drive into “The City”. “Living in the country only used to be for rich people”.
The expansion didn’t stop there with these same engines powering locomotives (railroads) which enabled even more expansion to the West. Before we had the Internet Superhighway and the American Highway system, every town in America was connected by railroads.
Sidebar - the America Highway System Creation (from Wikipedia)
In the 20th century, the United States Congress began funding roadways through the Federal Aid Road Act of 1916.
In 1926, the United States Numbered Highway System was established, creating the first national road numbering system for cross-country travel. United States Numbered Highways ranged from two-lane country roads to multi-lane freeways.
In 1953, the US govt developed a proposal for an interstate highway system, eventually resulting in the enactment of the Federal-Aid Highway Act of 1956.
The invention of the automobile in the early 1900s spawned the US highway system which in turn created an all new trucking industry followed by the rise of companies like FedEx and UPS. There’s Jevons at work again.
The US economy scaled massively during this time as the marginal cost of transportation incentivized all new industries, business models, and companies.
Instead of reducing the human jobs of riding and managing the horses, the new automobile industry created an exponential amount of jobs and was once the leading powerhouse of the entire US economy - based out of Detroit.
The same technological changes witnessed between horses and automobiles has been repeated over and over again with airplanes, radio, the invention of TV and the transistor.
Welcome to Jevons Paradox.
“As technological progress increases the efficiency with which a resource is used, the rate of consumption of that resource rises due to increasing demand.”
Jevons Paradox is what Bilal’s post is highlighting with his brilliant “The Cost of Intelligence Is Going To Zero”. He emphasized the historical example of the marginal cost of communication going to zero in just the last 25 years (roughly 1995 - 2020).
This ease and speed of communication created all new markets never even contemplated prior to the internet. We created the internet browser in the early 90s which spawned “the world wide web”, e-commerce, and then mobile phones. These new ways of communicating created millions of jobs and technology powerhouses such as Google, Facebook, WhatsApp, Netflix, and Zoom.
What history has actually taught us is that new technologies do eliminate “old jobs” but create multiples of brand new jobs from the new technologies, industries, and companies created.
Today we are living through the next big technology shift - maybe even the biggest one - with Ai. Likely to be larger and have more impact than the internet and the rise of the browser and the cloud.
Here’s a correlation equation for Jevons in more laypersons terms:
Faster = More Efficient
More Efficient = More Usage
More Usage = More Demand
More Demand = More Supply of New Companies and Jobs to Fulfill the Demand
To bring the point home, here are 3 more examples.
Example 1: The Spreadsheet Didn’t Replace Accountants
When VisiCalc and Lotus 1-2-3 arrived in the 1980s, people assumed they would eliminate the need for large accounting departments. Instead, spreadsheets became the engine of most everything financial.
The scope of finance exploded. We didn’t fire accountants. We asked them to do more. We created the FP&A industry. We forecasted faster, scenario planned better and the rise of investment banking and Goldman Sachs and Morgan Stanley and Blackrock were all born on the backs of the ”little ol’ spreadsheet that could”.
Example 2: The Telephone Didn’t Kill Travel
When Alexander Graham Bell invented the telephone in 1876 at the age of 29, the pundits of the day predicted that the railroads would disappear. After all, why travel across the country when you could just make a phone call?
The pundits and naysayers were wrong. Travel boomed shortly after this new amazing technology invention.
Why? Because the phone made it easier to initiate new relationships, but it didn’t connect people the way humans needed to connect. Telephones made the world smaller with faster more efficient communications but they ended up using Railroads more to go finish the business that may have started with the phone call.
Example 3: Search Engines Didn’t Eliminate Internet Usage
Google didn’t end the search for answers on the internet. It obviously, in hindsight, massively scaled it. And yet, a significant amount of “experts” were saying Google/Search Engines in the late 90s/early 2000s would eliminate the need for the majority of the recently created internet companies. HA!
Faster and easier access to knowledge led to more, more, more. More SEO consultants, content strategists, and the rise of the algorithm. Well, okay, there was a downside with that “rise” we’ve now all lived through due to the algorithmic overlords.
But the fact remains, faster and easier = more usage not less.
Enter AI:
We are now watching Jevons Paradox in real time with all things Ai.
Will Ai replace several current jobs? Absolutely, yes. But no different than the human pooper scoopers. But that’s the wrong question.
The right question is: What new industries, business models, and all new companies will be created that never existed before… because they couldn’t exist without these new Ai technologies?
Just as the automobile created the suburbs, Ai will create all new business models.
I believe we are vastly underestimating this upcoming shift.
So, let’s go back to bz’s post (Bilal) which “prompted” this post!
He outlines all these new jobs… here’s a screenshot from bz’s LinkedIn article making this crystal clear:
Now Enter Amara’s Law:
“We tend to overestimate the effect of a technology in the short run and underestimate the effect in the long run.”
Maybe the best Amara example was the Dot-com boom and bust in the late 90’s. Everyone thought this “new technology” would dominate in the short run. Then came the crash and many abandoned the idea of e-commerce and the internet itself as simply a bubble. 25 yrs later, it’s obvious it was not a bubble. We simply overestimated the speed of adoption in the short run and underestimated the impact of all things internet in the long run. Imagine a world today without Google, Amazon, Netflix, Instagram, and so many others that have transformed our economy and society in the “longer run”.
Today Ai can draft emails, summarize meetings, write code, and hallucinate a Shakespearean screenplay on demand.
But it’s not the tasks we’re automating today (short run) that matter most.
It’s the new questions we haven’t yet learned how to ask (long run) and the new systems we’ll need to build to govern what I’ve called our modern day Cambrian explosion of generative technological capability.
Prediction: The Future Will Be More Human, Not Less
Here’s my take:
Ai won’t replace operators… it will pressure-test our judgment… and separate the wheat from the chaff.
CFOs and COOs won’t be replaced by LLMs. Ai will significantly enhance our cognitive and influence capabilities. Our roles will shift more and more to designing and executing increasingly complex business model decision chains made by hybrid teams of humans and Ai agents.
The org chart will bifurcate into “Prompters - Analysts,” “Feedback-Loop-Coders”, and “Human+Ai Agents”.
Ai won’t replace your team… you just won’t need as big a team. Ai will upscale your team and pressure-test everyone’s thinking. Those that can’t upscale? Well, there will still be plenty of more traditional jobs requiring human dexterity, creative manual construction, and infrastructure building to support all this “thinking”.
The future won’t be less human. It will demand better human judgment and creativity to those that design systems, supervise Ai recommendations, and govern the scale of demand that is clearly coming. New industries and business models will be born which will spawn all new companies and all new jobs.
Once upon a time we were a farming society. That DNA of farmers became some of our best inventors (Leonardo da Vinci, Ben Franklin (the orig. polymath), Nikola Tesla, Henry Ford, Edison, Wright Brothers, Graham-Bell) - yes they all had rural, agricultural “roots”!
We’ve now planted new “seeds” called Ai… it’s time to sow more creativity and reap our future rewards.
Thanks Bilal for inspiring this post with your “Cost of Intelligence is Going to Zero” post and forcing me to connect Jevons and Ai here.