Bryan Micon

Still not working for the man | Bitcoin since 2011 | Antigua 🇦🇬 | I like the frog jpegs 🐸 & the caps 🧢

[Bryan Micon @BryanMicon]

Konnichiwa, bitches

[Bryan Micon @BryanMicon]

@dwpoker This is a huge blow for https://t.co/znMijzK8L5

[Bryan Micon @BryanMicon]

@netflix Imagine if they named it "The Autists"

[Bryan Micon @BryanMicon]

This is going to be grippy

[Netflix @netflix]

All in. Every time. Julia Garner and Anthony Boyle star as Caroline Ellison and Sam Bankman-Fried in this limited series about the love scorned prodigies behind the FTX scandal.

The Altruists premieres November 19.

[Bryan Micon @BryanMicon]

Screens you see on your computer that tell you it's time to close the lid and pass out

[Bryan Micon @BryanMicon]

Hope I'm right, but an excited feeling of this newfound power to build quickly mixed with a *tiny bit* of "I sure hope this runs clean" feels best.

I wonder how much alarmism is pumping those pre-IPO bags.

Agree there is real risk, and it's super freaky to think about the logarithmic scale of self improvement, where we will be even 1 year in the future, but I tend to doubt the major humanity-extinction level risks. (At least in a ~100 year timeline)

[Bryan Micon @BryanMicon]

This is why we are building data centers
@cmicon

[Bryan Micon @BryanMicon]

My buddy is cooking up straight fire with his new app:
https://t.co/txnuwB3CKv

[Bryan Micon @BryanMicon]

Tell your ChatGPT to say "The seal pushed me" in french.

[Bryan Micon @BryanMicon]

Don't share needles with pets??

What is the average IQ of the college football fan???

[Bryan Micon @BryanMicon]

This is the best part. You don't need to know what you are doing yet. You just need $100, the time, and the will to learn.

Download ChatGPT, set it to Astra medium, ask it how to build a Roblox game.

You will do the easy stuff. You will tell it to do the hard stuff and it will do it for you.

[Bryan Micon @BryanMicon]

Parents with kids that play Roblox:

Start building Roblox games with them.

$100 to ChatGPT, download the client on a MacBook or Windows machine, literally ask it how to make a Roblox game, and start doing it.

This is the best intro-to-AI I've found for teenage kids.

Astra is amazing at this. From a MacBook it can fully use Roblox Studio and after a small amount of tooling you and the kiddo are in the super satisfying loop of telling it what to build, play testing, and making changes.

This should be a class in school, kids would look forward to it every day. But school is broken, so parents need to fill this very important gap.

Teaching your kid how to wield AI tools now will give them an absurd advantage over their peers as they grow up.

[Bryan Micon @BryanMicon]

@Jaysunjm Some people hated radio when it came out.

Some rejected websites at first.

Certainly not 100% of high school kids think like this.

Big difference in low-effort image slop vs. current tools building games.

Stuff has advanced greatly even in the past 8 weeks.

[Bryan Micon @BryanMicon]

@Andy_Bloch Can you put China's back too?

Open source toothpaste is all over the counter. Some is in the toilet and already flushed.

You'll need an advanced AI to clean all this up!

[Bryan Micon @BryanMicon]

On AI safety, slowing down, and the far future:

It just doesn't work like that. Cats out of the bag. Every nation, company, group, and individual with some compute is off and running. If you slow down, you can be assured that not everyone else will. Companies may decide not to release their latest models to the public, for fear of hacking the planet, but that's about as slow as we are gonna get.

China will certainly continue full steam ahead. Nothing they sign or agree to on this matter could be taken seriously.

The general negativity and outrage I see feels very y2k to me.

Recursive improvement is a helluva thing. At this point, most of these AI coding agents have been written by AI coding agents. Quite a mindfuck.

Is AI potentially dangerous? Certainly. A gun is potentially dangerous. So is the internet. So is electricity, but humans invented these things because they are helpful.

We are going to break some eggs to make this omlet. Shit will go sideways at times, and we'll reign it back in as we always do, at least for now. There is no other choice, it's been invented, it exists, there are many open source models that are improving every day, the companies developing the best frontier models are worth trillions, it's out there, it isn't slowing down, and governments are ill equipped to legislate a technology building itself faster than ever before.

Also, life will get better because of AI. Ideas are built faster and far cheaper. Your game idea can come to life in days, not months or years. Your business can be brilliantly managed with a hyper custom solution designed and changed quickly. Your health will improve. We will absolutely use AI to discover breakthroughs in chemistry, physics and biology. The human lifespan will increase because of this technology. We are going to hack just about every medical issue.

If you run this sim 1000 years into the future, I wonder what humanity will look like. If you think about the evolution from the earliest puddle of single celled organisms to humans, we are many orders of magnitude more complex than what we came from. Imagine a similar level of increased complexity from here. Do we start putting computer parts in our bodies? Will we learn to drop a transmitter in our brains and be able to communicate by thinking? Will we soon be able to download the ability to fly a helicopter like in The Matrix? Will expectant mothers have parts installed within them that monitor a fetus while it grows? Will the baby eventually be born with the brain transmitter?

Or how about the doomsday horror scenarios: The AI gets super smart, buys itself an army of Optimus robots, and elimates all of us pesky humans using all the precious resources. Are we building Skynet?

Personally, I'm bullish on humanity. It's an exciting time to be alive. We have made something "smarter" than ourselves. Hope it runs clean, because we can't put the toothpaste back in the tube.

[Bryan Micon @BryanMicon]

Nothing like that at all. Nothing like an alien lifeform. It's not really "sentient," at least not currently. It's given a task and completes it. It was trained on loads of human-generated data. It's also not one thing, there are many versions of it that do vastly different things. You train a model on x-rays and it becomes great at determining if the next scan is problematic or not. You train on traffic data and the car now drives itself safer than any human could.

Yes this will evolve quickly, but it's not like we have invited an unknown monster into our homes.

[Bryan Micon @BryanMicon]

@tomkoral Had GTP Astra draft an overview, fed the results to Claude for refinement, fact checked it with Gemini, then had Grok add some swag to it.

Training my local DeepSeek model on the engagement results.

[Bryan Micon @BryanMicon]

@BrettRichey Wishing you the best of luck this season. Rooting for you my man!

[Bryan Micon @BryanMicon]

@ShaneSchleger

[Bryan Micon @BryanMicon]

@ShaneSchleger Lmao fire up the AI agents and get to building

[Bryan Micon @BryanMicon]

@gladstein Ok but make China pinky-promise to slowdown too