The AI We Need VS. The AI We Want

Total Factor Productivity Isn’t Everything, but in the long run it’s almost everything.
- Paul Krugman

There are two types of innovation: indulgence innovation and abundance innovation. 

When indulgence innovation is thriving, we spend more time glued to remote devices gaming, gambling, scrolling, and buying crap that we don’t need. We spend less time exercising, solving important problems, and connecting with others. For example, Facebook’s creation of the infinite scroll was an example of indulgence innovation. Some people and companies made lots of money from it. But society is arguably worse off because of it. 

When abundance innovation is thriving you feel it in the areas of your life that you seriously depend on. Healthcare is faster and better. Highways, schools and bridges are built faster, with less money, less corruption, and fewer people. Insurance claims are processed faster. The world transitions quickly from vehicles and infrastructure that’s dirty and inefficient to infrastructure which is faster and cleaner. For example, the record-breaking speed of discovering the COVID vaccine by BioNTech and Pfizer was great abundance innovation. So too was DARPA’s development of the internet. (Another interesting thing I realized while writing this is that prior to X (formerly Twitter) almost all of Elon Musk’s startups were in abundance fields.)

The question about AI is whether it will accelerate indulgence innovation or abundance innovation? Will it enable us to solve important problems where we’ve regressed? Or will it makes us more addicted to our screens? 

There’s an economic indicator that helps answer this question: total factor productivity. It basically measures how much more an economy produces with the same amount of money spent, hours worked, and energy consumed. When total factor productivity goes up it’s a good sign of innovation progress. When it slows down it’s a sign of stagnation. And in the United States, despite the ascent of the internet, social media, and now AI, total factor productivity has been nearly flat since 2000. We’ve been innovating a lot and building nothing, as far as this innovation barometer is concerned. Many of the great innovations of the past two decades make us less productive. Conversely, many of the products and services that are essential to our lives are slower, more expensive, and less reliable. The economist Tyler Cowen calls the 30-year slowdown of total factor productivity the Great Stagnation.

The reason total factor productivity has been slowing down in America during a time of seemingly explosive innovation is because we’ve been innovating in indulgence, and stagnating in abundance.  Total factor productivity has quintupled over the past decade in gaming, media, telecommunications and e-commerce. Conversely, it is **negative** in transportation, investment banking, insurance, and construction. Meaning, it takes more money, people, energy, and time to produce the same outcomes in these industries than it did 15 years ago.

Will AI Make Us Innovate Better Or Worse?

Whether AI will make innovation better or worse from a total factor productivity perspective depends on who you talk to.

If you speak to Rajvir Madan, the Chief Digital Officer of Arcutis Therapeutics you start to feel optimistic about the innovation impact of AI. Using AI he has been able to reduce the drug discovery and development cycle from 12 years to 9 months. AI enables biochemists to discover patterns between patient data and molecular data faster, and smarter than people can. This might help revive total factor productivity in healthcare which has declined nearly 3% over the past decade.

Raj has another advantage. He’s part of the Entrepreneur Exodus. He left a stable, prestigious, senior job at GSK to become an entrepreneur when he realized he could be faster, and more impactful if he started from scratch. He’s not alone in this regard. A record 5.5 million Americans have left leadership in large corporations to launch new ventures over the past 5 years. Many of them are deploying AI in systemically important, stagnated industries. 

Most of the Large Language Models like Chat GPT, Bard, and Anthropic were built on content that fueled innovation indulgence, and made total factor productivity stagnate. I think their continued growth will perpetuate the Great Stagnation.

But, AI is driving incredible breakthroughs in areas like healthcare, transportation, and insurance, which have stagnated for years. Thousands of brilliant innovators are leaving the companies which perpetuated the stagnation and making incredible breakthroughs with AI once they leave. The industries are less sexy. The entrepreneurs tend to be a little older. But their impact is cause for optimism. 

Here’s the simple math of why I feel optimistic:

• There are 348 AI Unicorns in the United States.

• 145 of those startups are in industries which are stagnating, and systemically significant.

• 62 of these startups are led by CEOs who once worked in one of the incumbents in the same industry where they now operate. 

I believe they have a real opportunity to create systemically significant innovation in industries that make our lives and economy better.