What is next
If you know me, you know I took a career break. Hard to believe it has been a year. Since then I’ve spent a lot of time trying to figure out what’s next. I’ve considered starting a company. Buying a company. I’ve tried a few ideas here and there. But, I keep getting stuck on one question.
What is next? What are the second order effects of LLMs?
I cannot get this out of my head.
Throughout history, every technological breakthrough creates a wave of change. Sometimes that wave wipes out entire industries. Sometimes it creates even more work than existed before. Sometimes it is merely a ripple. At first glance those outcomes seem random.
When henry ford made the automobile, what happened to horses? Specifically, what happened to all of the people whose livelihoods revolved around horses? Stable hands, blacksmiths, and carriage builders were put out of business.
But, the interesting part isn’t what disappeared but what appeared. Someone had to build and manage gas stations. Someone had to pave roads. Roadside hotels were not new, but they were seeing traffic like never before. Fast food, road maps, the list is exhaustive but all of these businesses only make sense as people adopt automobiles.
If you tried to get into the automobile industry you were doomed to fail. It was capital intensive, labor intensive, and required specialized skill sets. But, starting or expanding a hotel? Building a gas station? Easy (in comparison).
The tool pivoted demand, rendering supply valueless.
Compare that to something like excel or double entry ledger accounting.
You would think spreadsheets would have devastated accounting. Instead there are more accountants today than ever. Excel automated away enormous amounts of tedious work, but instead of eliminating accountants, it made accounting cheaper. And since it was cheaper, more companies could afford to do it. Which then made financial modeling accessible, which then expanded auditing, and future forecasting became commonplace.
The tool increased supply, but demand increased even faster.
History is similarly littered with false starts. Times where it seemed like a technology would be transformative, but it never really took off. Things like segways which could have transformed cities, but never did. Only to later see cities transform because of electric scooters and bikes. Or the metaverse replacing offices, only to later see a pandemic push people towards similar but different offerings. None of the technologies disappeared, and its not to say that the technology didnt shape society, but they didn’t fundamentally reshape society.
Sometimes a technology is impressive, but not transformative.
It is somewhat fascinating, because the automobile and excel technologies seem similar on the surface. Both automated work inside an existing profession (sound familiar to LLMs?). One decimated an ecosystem, the other expanded it.
Traditional economics teaches that when supply increases prices fall. However, supply side economics teaches a different notion. Sometimes lowering the cost of something creates an explosion in demand instead and we see it all over the place.
Space launch costs have fallen dramatically, and suddenly people are talking about building entire businesses in orbit. You have pharmaceuticals being manufactured in space now!
Sometimes making something cheaper doesnt shrink the market, it makes it bigger.
Ive spent most of my career chasing that phenomenon to an extent.
I went to school for VLSI, designing transistors and integrated circuits. Fascinating work, but a relatively small industry being increasingly automated.
So I moved up the stack and into embedded systems where there are far more opportunities. But, I quickly realized embedded systems were not growing nearly as quickly as other industries and moved into software and cloud engineering. So, I leveraged my skills and followed that wave. As an aside, its funny to me how similar all three are, but perhaps thats another post.
There are a billion posts about how LLMs will replace software engineers and developers. Everyone agrees that LLMs are getting better at writing code. People disagree on the extent to which LLMs can replace different disciplines. Whether software engineers will become stable hands or accountants is a moot point to my metapoint.
It brings me back to the question I cannot stop thinking about. What are the second order effects of an LLM? What is next?
Most people are focused on the first order effects of LLMs. AI writing code. AI replacing customer support. VC firms are pouring in billions and people are quitting their jobs to build a dozen variations of the same thing in this space.
But, to me, the automobile wasnt interesting because it replaced horses. It was interesting because it created industries that nobody had imagined. It expanded some (roadside hotels) while decimating others (hay prices).
Can I identify one of these opportunities before everyone else? Can I take advantage of it? Or, are those second order effects only obvious in hindsight. Its the question I cannot get out of my head.
** Its kinda funny when you think about it. The person who made the “shovel” for the automobile, oil is rockefeller who is largely one of the first ultra rich individuals ever. Probably explains NVIDIAs insane market cap.