To Zero and Back Again: The Coming Revolution In AI

“Perhaps the most important consequence of the Internet, at least from a business perspective, [is] the reduction of the cost of distribution to effectively zero.”  

 -- Stratechery

Technological revolutions tend to wreak havoc on old business models. In 1900, for example, there were only 8,000 automobiles in the U.S. At that time, autos were expensive, slow and unreliable, and most people chose to travel by horse or train. But after Ford pioneered mass production and steel displaced iron as the engineering material of choice, things changed radically. After 1900, auto production doubled every couple of years, and by 1920 there was one auto for every 11 persons in the U.S. By 1930, six times more people traveled by auto than by railroad.

Technological change is not random, nor is it a process of constant, linear progress and cumulative development. Rather, for the past 250 years, technological change has occurred in the form of revolutions. In her book “Technological Revolutions and Financial Capital, “ Carlota Perez defines a technological revolution as “a powerful and highly visible cluster of new and dynamic technologies, products and industries, capable of bringing about an upheaval in the whole fabric of the economy.”

Anyone working in media since the late ‘90s has felt many such upheavals. From music to publishing to radio to TV, the Internet has transformed, unbundled and reshaped the entire media landscape, a process which continues to this day. 

Historically, the media industry has had a lock on two things: content creation, and content distribution. During television’s Golden Age, for example, NBC, CBS and ABC had total control over both creation and distribution, and thus monopolized consumer attention. Cable TV altered that equation, and the Internet blew it up forever, bringing the cost of distribution to near-zero (while the cost of creating good content has proven more tenacious; witness Netflix’s $6 billion content production budget in 2017). 

Should we have been surprised? The Internet, after all, wasn’t just some thing that appeared out of nowhere, like Athena sprung from the head of Zeus. It was the coalescence of radical advances across set of technologies – microprocessors, computer hardware, computer software, telecommunications -- that had been in development for two decades and fueled by massive capital investment (public and private).

Back in the 2000s, I spent time in both the music and publishing industries, and despite being surrounded by supremely talented and creative people, the mixture of surprise and dread that suffused the environment never seemed to dissipate.

Perhaps these things are easier to see in hindsight than when you’re in the thick of them. Technological revolutions tend to have a life cycle of about 50 years, meaning that many adults can spend much of their working lives thinking about and acquiring skills relevant to the last revolution —not the next one. But if history is any guide, there’s always a next one.

There are strong indications that the next technological revolution will be artificial intelligence.

For decades, AI has been a sci-fi dream, but core developments in several complementary areas are now colliding to make AI a reality: 

-- Huge improvements in processing speed and architecture (CPUs become GPUs).

-- Significant advancement in the underlying science as well as specific applications, such as neural networks.

-- Access to massive, Internet-enabled data sets that can be used for reinforcement learning.

-- Huge capital investment—not just from VCs, but large corporations like Google and Facebook.

Rather than describe all the ways specific AI applications are already changing our lives (Amazon Echo, bots), now is the time to ask the big questions. Not the Skynet-meets-Matrix dystopian questions, but the ones likely to be relevant to individuals and businesses in the next decade. What business models will go to zero when AI becomes mainstream?

Stanford’s “One Hundred Year Study on Artificial Intelligence” provides some clues. According to the study, the industries likely to be significantly reshaped by AI in the coming years include transportation, healthcare, education, finance, and entertainment, among others.

When an AI can learn to drive a car and diagnose disease better than a human, what skills should a human learn?

Because there are no clear answers to any of these questions yet doesn’t mean we shouldn’t be asking them. We know from history and recent experience that the industries most negatively affected by new technologies tend to be the ones that ignore the impermanence of the current paradigm. This is a failure of both imagination and analysis.

Every technological revolution brings with it more jobs, and more value creation, than the one that preceded it. And that should be good news, even if it means working side-by-side with a robot.