Stratechery on the first wave of AI

Oftentimes when new technological advances happen, companies are focused on covering new marginal costs rather than (as they often do later) focusing on how low-marginal-cost might drive more business.

Tech has three philosophies.

Google & Facebook: computers help you get things done by doing things for you. They're aggregators which attract people by their usefulness.

Apple & Microsoft: the computer enables you to do your work better and more efficiently. Computer are an aid to humans and not their replacement.

(Salesforce) tech’s third philosophy: improving the bottom line for large enterprises.

The cost of change management for these tools is one of the central problems. The personal computer wasn't adopted by making it dumber (MSFT clippy) but by new generations growing up with them as a natural extension of who they are.

Or, to put it in rather more dire terms, the initial value in computing wasn’t created by helping Boomers do their job more efficiently, but rather by replacing entire swathes of them completely.

Agents aren’t copilots; they are replacements. They do work in place of humans

Salesforce isn't aiming to make people more productive but they're aiming to make companies more productive. Similar to mainframes: they replaced people while giving managers information.

This is manageable as long as a motivated human is in the loop; what seems unlikely to me is any sort of autonomous agent actually operating in a way that makes a company more efficient without an extensive amount of oversight that ends up making the entire endeavor more expensive.

LLMs are path dependent and once they take an approach they have a weakness known as “auto-regressive large language models”.

While o1 is slow and expensive for humans to use, in an agentic capacity model, as a replacement for humans, it would make sense because it's scalable.

Better data still results in better outputs.

To that end, the company I am most intrigued by, for what I think will be the first wave of AI, is Palantir.

Palantir is akin to the mainframe.

We are already seeing that that is the case in terms of user behavior: the relationship of most employees to AI is like the relationship of most corporate employees to PCs in the 1980s; sure, they’ll use it if they have to, but they don’t want to transform how they work. That will fall on the next generation.

The first wave of AI will likely replace humans and not focus on making them more efficient... just like with computing.

My core contention here, however, is that AI truly is a new way of computing... Transformers are the transistor, and mainframes are today’s models.

True consumerization of AI will be left to the next generation who will have never known a world without it.