Welcome back to the AI For Humans newsletter!

No, seriously, welcome back. It’s been a while since we posted. And, lucky us, a lot of you subscribed while we were gone. If it’s your first time, we’ll be sending these out weekly (on Mondays) and try to focus on the sorts of AI analysis, apps, and tools that might be meaningful to you, the human.

What happened? Where did we go? Well, the short hiatus from the newsletter for the last few months was in service of our very exciting news:

We’ve got a new company called AndThen!

You can learn more about AndThen at the end of today’s post, and sincere thanks to everyone for supporting us over the last few years. But first…

Does GPT-5 ‘Sucking’ Mean We’re in an AI Bubble??

Hold your horses there, partner. The ‘suckiness’ of GPT-5 is vastly overrated.

GPT-5 is a fine new AI model. Yes, it’s a bit more finicky to prompt & there were some significant kinks to work out, but GPT-5 Thinking mode is currently the best reasoning AI on the market. And GPT-5 Pro is even better if you’re willing to shell out the $200 bucks a month.

However, GPT-5 is not a significant leap. It’s a step past what we had but not a major leap up (say 30%+ improvement) in capabilities. Just a few months ago, I posted this video on YouTube about what it might mean if GPT-5 wasn’t the end-all, be-all and here we are.

However, the death of AI has been vastly overstated.

But Sam Altman Himself Said It Was A Bubble! And What About That MIT Study?

In case you missed it, Sam hosted an on-the-record dinner with journalists after the launch of GPT-5 to talk about what went wrong with the GPT-5 launch.

And then he said this:

This set off alarm bells across the tech and financial sectors, and, combined with the widely read MIT report that says (along with MANY other things) “95% of organizations are getting zero return” on their AI investments, things got very dark in the AI world.

But there’s a whole different part of the exact same MIT report that says the following:

The much more important part of this story is that you, the normal human, are finding AI tools useful and making your life vastly easier and better.

It’s just not through ‘official’ channels. In other words, the CEO might not know what to do with AI tools but you actually might.

The Transition From New To Useful To Economically Useful Takes Time

A useful metaphor for thinking about this comes to us from history.

Lots of people have compared generative AI to the invention and rise of electricity. Which is somewhat ironic considering how much electricity it takes to power AI.

However, electricity to a very long time to reach the average human. The wiring had to be installed, power lines erected, additional products had to be invented to use it.

Eventually, it transformed our entire society. But that happened over time as we built the stuff needed to make the best use of it.

ChatGPT alone, because of the distribution curve of the internet, is currently being used by 700m people a week and it launched less than 3 years ago. And as soon as a new tool rolls out, everyone gets instant access to it.

It makes sense that none of us have figured out what to do with it yet.

GPT-5 Being ‘Underwhelming’ Is A Gift. Use It.

A flat release buys us, as humans, breathing room. Super intelligence might be a few more years away and, honestly? Great. We needed more time.

So here’s the move. Stop waiting for a miracle model and build one small habit. Pick a task you touch every week and turn it into a tiny, repeatable workflow. Write the prompt, save the steps, run it twice. The moment you can do it without thinking, you’ll feel the AGI. That is how hype becomes habit, and habit turns into real value.

More on Thursday!

- Gavin (and Kevin)

THIS WEEK’S AI FOR HUMANS: Is Google Set Up To Overtake OpenAI? 👇

We 💛 This: AndThen is our new company!

This is slightly selfish but we’re going to hijack this section to let you know that we’ve founded a new company and taken pre-seed funding from a16z Speedrun.

We just opened the early list for AndThen, and, while we’re not talking a ton about it yet, I can tell you in involves new & creative ways to interact with AI audio. For now.

Our ambitions are MUCH bigger but since we’re building on the edge, this is where we’re starting. And we’re having a ton of fun so far!

We’re huge believers in human creativity and that the best AI media formats are likely something entirely new we haven’t seen before. That’s the space we’re working in.

I shared a little context on my LinkedIn. Add your email now to be first in line when we drop the preview.