Why Winterspeak 2.0?
"Generative AI will automate tasks just like the Internet reduced friction"
I sat across from a friend of mine over dinner. It was the late 90s. “The internet reduces friction,” he said, “that’s how to think about it.”
He wasn’t wrong, but he wasn’t right either. The internet does reduce friction, but friction defines whether we pick choice A or choice B, it determines the boundaries of the firm, it shapes behavior and structure. Reducing friction doesn’t mean “same as now, just faster”. It means different.
GenAI is the same. Yes, we can automate tasks, but this automation will similarly impact behavior and structure and lead to something different. The purpose of this newsletter is to work out what those impacts are.
My theses:
If the Internet opened up a new frontier (see “Homesteading the Noosphere”), and the first generation of ML transformed the demand side of the digital economy through ranking (algorithmic recommendations), then the second generation of ML, generative AI, is going to transform the supply side.
We will have 10x the creators we do now. Each will generate 10x the output. The total volume of content will be two orders of magnitude greater than what we have currently.
Since each individual can be more productive, team sizes will shrink, as diseconomies of scale dominate. People will bring more general skills.
Incumbents will not be able to manage this scale transition. They will be outcompeted by entrants who just won’t hire people to begin with (major caveats below).
Caveat 1: With digital content being commoditized, the complements (distribution, physical connections, IRL, actual people, beauty, etc.) will become more important. Many businesses have material non-digital components, and they will benefit.
Caveat 2: Digital distribution platforms may become larger, as ranking becomes ever more important.
Caveat 3: There is a jagged edge in terms of what AI can and cannot do (ie. it’s difficult to predict what will work and what will not) and by extension it’s difficult to predict who makes it through an AI world and who doesn’t.
The corollary to all of this is innovation will start at the micro-level, with users applying it to local problems. AI-first scale solutions will grow from those findings.
We will need decentralized moderation and dispute resolution systems to handle this volume.
We will need better signals of quality to match people with the right content.
This project is to dive into micro-innovation at the workflow level, and then see how that impacts first organizational design, then industry structure, and ultimately the overall economy.