Stop blackpilling, stop dooming, forget p(doom), and learn to love building GPT wrappers
A model of development plans in Phu Quoc, Vietnam
In the south of Vietnam there is an island off the coast called Phu Quoc. For years not many people lived there and not much happened. It had beautiful beaches and some small fishing villages. There used to be an airport the size of a small interstate motel.
Then the Kien Binh - Phu Quoc 110kV transmission line was built. The longest sea-crossing submarine cable in Southeast Asia (at 57 km) was built in 2014.
Since then the island has been transformed. It has become filled with resorts, amusement parks, and luxury residential real estate. A new airport was built and the government made the island visa free for 30 days (the rest of Vietnam typically requires an e-visa even for short tourist stays).
All it took was power and capital.
The point of this introduction is to tie together a couple of things:
- Asia is building things and the speed at which it’s happening is incredible.
- The west has not been and as a kid who grew up in the 90s it feels like the west has horrifically decayed in the last 25 years.
- AI is what we need to kick us into building again.
- If you are blackpilled and p(doom) obsessed stop. The rest of the world is going to move on without you.
Hyperstition
Hyperstition, a term coined by Nick Land and Mark Fisher, describes ideas or narratives that become real through their widespread belief and dissemination. It’s a blend of “hyper” (beyond or above) and “superstition,” suggesting a belief system that, through its very existence, shapes reality. Essentially, hyperstitions are self-fulfilling prophecies or beliefs that, when widely adopted, can cause their own manifestation
By blackpilling you are making bad outcomes more likely.
By focusing on p(doom) and trying to erect regulatory barriers you are making it more likely that some megacorp controls the most powerful AI system and open source and decentralized systems never get off the ground.
By talking about bad outcomes obsessively you are not going to be able to build anything.
If you build nothing and give up all power you ensure that you have no control over the outcomes of our possible futures.
Just relax and build GPT wrappers
LLMs are such a generally useful new paradigm in the capabilities of computers you need to stop and rethink what you can do.
Don’t look at your product you designed in 2010 and think about how you can “add AI to it”.
Start from first principles and think about how you could solve the same problems your product solves but with a totally new and novel set of capabilities.
Openrouter token usage June 2025
Negative p(doom)
I listened to Roko Mijic on Liron’s Shapira’s interview show Doom Debates recently. Roko argued his p(doom) is negative 60%.
Meaning: western civilization is screwed for a number of reasons and is experiencing a rapid decline in many measurable metrics. If we don’t get AI it’s almost certainly destined to be over. If we do get AI, we have a 60% probability of pulling through it and creating a new golden age.
I think this makes sense.
There is something about the power production capabilities of a civilization and its flourishing that are inextricably linked.
Right now, AI is the thing that will drive us to produce more power and increase our quality of life, and help us move past the malaise and downward spiral the entire western hemisphere is trapped in.
You can just stop blackpilling and start building the AI-powered future one token at a time.
Go out and consume tokens.