What this Substack is about
Note: I started this Substack in 2025 while diving into the field of AI safety. My perspective and approach have evolved, broadened, and complexified since then, and those changes will soon be reflected in this About page.
My goal with this Substack is to help capable people (like yourself!) better understand the range of threats that AI could pose.
I hope that if enough people are informed of what’s going on and what’s at stake, we can collectively increase the chances that things go well.
My background is in content marketing, and I’m seeking to transition into full-time AI safety comms.
A bit about me
Hey 👋 I’m Mike.
I studied mathematics at Dartmouth back in 2013. (I know the Ivy League name drop is cringey. But I’m hoping you’ll trust that I’m at least not a complete dunce!)
Then I spent most of my career in the personal development industry.
I have reverence for the power and potential of technology, and follow developments in tech and AI. And frankly, I’m excited about what AI could bring. I genuinely believe that AI could help us with our biggest societal and biological problems.
Though I don’t assume that technology is inherently good and beneficial. And if unsteered, I think it could cause tremendous harm.
And just to clean up one quick misconception: you might think that the people nervous of AI are just paranoid alarmists. Like those who were afraid of Y2K or the Mayan end of the world in 2012. But truthfully, I’m not an especially neurotic person who overreacts to potential danger. In fact, my top hobby is parkour, which involves jumping over high stuff.
(That’s me above!)
In other words, I’m not not anti-tech, nor someone who is constantly afraid of what bad things could happen.
But after a few “oh shit” moments, I’ve come to believe that we need to take the risks of AI way more seriously.
One of those moments was reading an article called AI 2027, where some respected AI researchers predict a hypothetical timeline of how things could go. And it’s horrifying.
Then I read The Precipice by Toby Ord. And that’s when I realized that it’s actually possible for humanity to destroy itself. And if we do, AI is probably how we’ll do it.
Since then, I’ve been drinking from the firehouse. And the more I researched, the more I saw that building AI is an insanely risky endeavor.
Humans losing control over a more powerful AI isn’t the only meaningful threat.
Whatever company or country builds human-level AI first could gain a dangerous amount of power.
Terrorists could use AI to create bioweapons.
We may build a society so reliant on AI that humanity becomes entirely disempowered.
Our young generations will grow up with chatbot friends, and we have no idea what that means.
Companies are plowing recklessly forward. There’s no national regulation. The risks are enormous.
I’m convinced of this to the point where I’m shifting my career into AI safety. I think this is a very big deal, and I want to spread awareness as best I can.
If you’d like to stay up to date with my reflections on AI safety and what you can do to help, subscribe below.
