How Freemium Almost Killed My Business, If You Want to Build, You Have to Believe, and More!

Hello, and happy Sunday!

We’re experimenting with a new format for our Sunday Digest—one that shares some of the things that caught our eye over the past week. You can find a roundup of everything we published in the latter half. Let us know what you think in the comments.

Reminder: Early-bird pricing ends tomorrow for our "How to Build an AI Chatbot" course. Learn to build in AI and save $500 today.

---

What matters this week in AI

President Biden signed an executive order mandating security requirements for companies training large next-generation models. It only applies to companies training frontier models larger than GPT-4, but it’s the first salvo in what will be a significant issue for the next decade. 

The executive order underscores an important question: how much is AI doomsday hype—and resulting government action—just cynical regulatory capture by large incumbents? Sam Altman says he isn’t doing regulatory capture. Ben Thompson disagrees:

“The point is this: if you accept the premise that regulation locks in incumbents, then it sure is notable that the early AI winners seem the most invested in generating alarm in Washington, D.C. about AI. This despite the fact that their concern is apparently not sufficiently high to, you know, stop their work. No, they are the responsible ones, the ones who care enough to call for regulation; all the better if concerns about imagined harms kneecap inevitable competitors.”

But AI doomsday can’t all be cynical plays by big companies to protect profits. Eliezier Yudkowsky has been writing about AI doomsday since the early aughts. OpenAI was created with safety as part of its charter. And not everyone in positions of power in tech is advocating for this executive order. A group of researchers and founders, including Meta’s Yann Lecun and Y Combinator CEO Garry Tan, signed an open letter arguing that it could stifle competition. 

Dan’s take

It’s right to point out the incentives of large players. But smart AI regulation that doesn’t hamstring startups is a good idea—arguing over who’s being cynical and who isn’t is pointless. Everyone in this debate has incentives. It’s better to argue about what kinds of regulations are necessary than trying to determine who has pure motives.

Chain of links ⛓️

AI isn’t a creative shortcut—it helps you build more ambitious projects.

AI turns everyone into a director or a producer.

The Beatles released a new single generated by AI, 50 years after their breakup. In AI for biology, you get what you train for.

New nonprofit Future House is building semi-autonomous AI agents for scientific discovery.

A preview of how AI could change job interviews (this one's fake, but it may not be for long).

One important product launch

Dot, an AI-first mobile operating system, is sort of like ChatGPT re-envisioned by Bret Victor. I’ve been hearing about it for months—friends of mine said it was the most impressive AI product they’d ever seen. I’m getting an in-person demo in San Francisco next week, so stay tuned. 

Dollars and data

At the trial of lapsed effective altruist Sam Bankman-Fried—who was found guilty of seven counts of fraud for his role in the collapse of FTX—Going Infinite author Michael Lewis sat near Bankman-Fried’s family and friends, not with the press.

The valuation of Twitter—er, X—dropped 55% to $19 billion.

Pinterest, whose stock rose 19% after earnings, is due for a resurgence.

Private equity sees blood in the water (paywall) for creator economy companies, which have borne the burden of missed expectations.

Now, on to what we published last week.

How Freemium Almost Killed My Business

What do you do when what you are “supposed” to do fails? That’s the question Bobby tackles in this essay. When a popular SaaS software growth methodology nearly kills his venture-backed startup he has to pivot or go out of business. He is honest, transparent, and articulate, and it’s relevant even if you don’t run a SaaS company. Read it so you can grapple with this question, “What strategy am I using just because I’m supposed to?”

Read

If You Want to Build, You Have to Believe

There are three forbidden topics in Silicon Valley: Elon Musk’s hair plugs, whether you can have work-life balance when building a startup, and God. Simone went with the easiest option and wrote about divinity. He argues that faith is an absolutely necessary and criminally under-discussed component of entrepreneurship. When you’re alone, grinding in the middle of the night, and money is in scarce supply, the only thing you can rely on is belief in yourself and your mission. Read this if you want to know upon which emotional well to draw when building gets hard.

Read

How to Win Arguments and Manipulate Managers

Behold! Evan, the valiant data-minstrel, dost challenge the notion that mere numbers hold sway in the hallowed halls of managers. With a jest and jape, he telleth of his own grand transformation from a humble scrivener of digits to a masterful weaver of stratagems, capable of bending even the most steadfast of barons to his will. His triad of artful tactics: to spin tales from tallies, divine the inner workings of the managerial mind, and brandish a trove of cunning ripostes for any parry. With the mirth of a jester and the insight of a sage, Armstrong teacheth how to transmute leaden data into golden narrative, awaken the slumbering dragon of empathy, and soar in stature within the noble court of employ. Thou must readest this text if ye desire a mighty and powerful skill set to manage bumbling kings. 

Read

Kate Lee Joins Every as Editor-in-Chief

The mark of a good hire is being congratulated by your friends. The mark of a great hire is everyone asking, “Why on earth did she join?” We are pleased to announce Kate Lee is in the second category. She’s been a book agent and the head of Stripe Press, and now, somehow, she is Every’s editor-in-chief and general manager. Over the coming months, as you see this newsletter improve, you can give all the credit to her. Read this to learn more about the future of Every and why Kate was excited to join.

Read

Is Chat the Future of AI?

Have you ever seen a dog so ugly it was cute? (Note from Evan: This is how my wife describes me.) Sometimes an object’s flaws make it that much better. So, too, with ChatGPT. Chat as an interface is inexact, imprecise, and prone to errors—and Dan argues that these are the things that actually make it great. A chat interface forces you to explore multiple paths of interest, enhancing your creativity. Read this if you feel like you aren’t getting enough out of ChatGPT. We promise it is more powerful than you think. 

Read