7 Comments

On the Ben Thompson essay: my impression is that he is trying to draw a distinction between the so-called 'bottoms up' customer acquistion strategy and the 'top down' customer acquisition strategy.

I'm not sure why he doesn't mention it in his essay, but the canonical example of a company which acquired its earliest customers from the bottoms up strategy is Stripe. Stripe got developers at all sorts of companies to use its simple code, they built stuff with it, told their managers that it was amazing, those managers told their managers, etc., and soon enough, Stripe penetrated thousands of organizations.

Compare, on the other hand, to, say, Oracle's or Salesforce's customer acquisition strategy, in which the sale was made, ideally to the CxO of a large enterprise. The actual users of the product (the front line employees) didn't have much choice in the matter of software they were to use to do their jobs; the software, often awkwardly designed, was foisted on them from on high. From the perspective of CxO types, this was great. From the perspective of the ultimate user of the software, this was terrible.

The question with AI is whether customer acquisition will follow the bottoms up or top down path. Ethan Mollick has written frequently in favor of the bottoms up approach, in which individual employees experiment with ChatGPT or similar, and figure out how to make it a productivity-enhancing complement. He has also noted that many corporate liability policies militate against employee adoption of AI tech.

Expand full comment

It makes sense the AI has much faster adoption than PCs or the internet. AI is free, requires no expertise, and requires no work to set up. You had to buy and install a modem and pay for a service to get the internet. You had to buy and learn a PC. AI just works. It's kind of like a new show on TV. It requires nothing on the users part other than knowing it exists and is good.

Expand full comment

My parents spent i think around $1500 for a good but not great Compaq Presario circa 1996 for our first PC. Not cheap! It was obsolete within about three years, also.

Expand full comment

"My sense is that people are very unmotivated to get personal coaching. It would be interesting to speculate on why that is."

The expansion of home and auto repair insurance suggests a deeper problem.

Expand full comment

The expansion of home and auto repair insurance might be downstream of the evolution of health insurance into pre-paid/subsidized healthcare.

Expand full comment

We need AI coaches—especially the poor homeless folk.

The govt should provide lower cost smartphones and track what the homeless folk do, with their money, with their physical bodies.

All govt aid should be going with increased monitoring of those spending the aid. Including all govt bureaucrats.

Expand full comment

Arnold should publish his speculations, on getting a coach, and other thoughts. My reason is that I get told enough what to do by my wife, after decades of being told by various bosses. I already have a long list of things I Should do … for a Better Life. Bah. I’m kinda mostly happy most days as is, and life is both getting things done and enjoying what you’re doing. Why doesn’t Arnold try an ai”Coach?

I like doing what I like when I like, because I like it then.

I’d also be interested in what he thinks Ben Thompson is saying. The message I got is that ai is a tech revolution … in processs … still missing something. Maybe the GUI.

Ben’s a bit wrong there, the User Interface is language, just like my interface to all readers here. It will be like HAL 9000, talking & writing & pictures. The LLMs are the early UI.

“Hal, open the pod bay doors.”

What’s missing is the LLM which probabilistically understands what we say we want, with command over other digital asst. agents to Do The Work, and do it correctly/ like I want. The early assistants that are usable include writing, and especially reading & summarizing books or long articles. Self-driving taxis for drunks will likely be here before fusion—tho ai controlled magnets for fusion is a big step, too.

Everything your computer guru friend can do, and do for you, a digiElf will soon be able to do.

Lots of mediocre coders will see no more promotions, and an increase in layoffs as good coders use Copilots to get lots more good coding done, and tested. Ben is right that Big Business decision makers want to save money & headcount, now.

He’s not quite saying what he makes me think: the CEO will want cost savings, layoffs, the smart productive bottom folk will be able to a lot more, there will a wave of middle managers laying off a lot of less productive folk. And then middle managers will mostly stop seeing promotions but instead see layoffs.

Meanwhile, customers will see lower prices, raised by govt inflation. Those keeping their jobs will be happier.

Expand full comment