The smartest move fomo made this year tbh
I’ve joined the @fomo team to lead partnerships.
I’ve been in crypto for nine years and have been thinking about crypto social for much of that time. I was Farcaster user #303, built the social trading network @clicker, and angel invested in fomo.
Then I ran my public fomo account (@kyle) up to top 20 all time. Now I get to help grow the product I’ve spent so much time using, and there's nowhere else I’d rather be.
I’ve admired @paulerlanger, @seyong, @qrazhan_ultd, and the team since they launched, and my conviction has only grown since joining. We have not seen a new social network break out since TikTok, and I believe fomo will be the next.
My focus at fomo is chains and team tags. If you want more activity on your chain or you want to reach your token holders directly, let’s talk.
Copypasting prompts is the new download button. Even open source repos feel a bit unnecessary at this point
inference retweeted
AI is going to know more about you than Facebook & Google combined
And we deserve the right to confide in a silicon intelligence that isn’t in a data center owned by a MegaCorp
Next Saturday (Nov 7) @DistributedMarz and I are hosting LOCALHOST, a Sovereign AI hackathon in NYC
inference retweeted
1/ As consumer agents start to take off with Muse, Instinct, and others to come
We wanted to build a tool that would help researchers understand what’s actually happening under the hood
We were especially motivated by reports of Instinct losing users’ money
Here’s a demo of our Interceptor tool with Muse:
inference retweeted
In last 24 hours, we gave the fly $10,000 to trade stocks on Robinhood chain. And simultaneously ran 2 simulations of random trades and just buy all available stocks and hold. robinhood:0x4eb990547bce4a982432ca88cf5fae7eed1a2d35
And the fly outperforms the other scenarios by a mile.
How the fly trades:
A real 165,122-neuron fly brain walks a dark room with five identical cards: TSLA, META, INTC, GOOGL, CRCL. It can't read the tickers. It smells the names.
It only trades when it stops on a card. Its memory centre leans toward or away from that stock compared with the others it just saw. Leans in → buys. Leans away from one it owns → sells. How hard it leans sets how much.
Holding isn't a decision. It's what happens when the fly doesn't stop.
When it walks back onto a stock it owns, price up is sugar and price down is a shock. That's how it learns.
The dashboard and the fly wallet will go live soon, and the fly is all set to be the richest fly in the world!
inference retweeted
Introducing Flyonardo da Vinci. The fly with a paint brush and a canvas.
A real fruit fly brain, 165122 neurons, the actual wiring, has a pen. Robinhood Chain is its senses: gas, whales, new launches, robinhood:0x4eb990547bce4a982432ca88cf5fae7eed1a2d35 buys and sells all fire real neurons, and 1,314 descending neurons steer the line. Every ~120 px, the next block's hash picks one of four colours.
Nobody steers it. Nothing is random. The chain picks the colour, the fly picks the line.
Each canvas keeps growing until someone claims it as an NFT for 10,000 robinhood:0x4eb990547bce4a982432ca88cf5fae7eed1a2d35, burned, not paid to anyone. The picture is pinned on-chain before it can be claimed, never chosen after.
Canvas #1 is already gone: 17,995 strokes, minted this morning. Canvas #2 is being painted right now.
Watch it draw: flybrain.online/flyonardo
github.com/fruitflydev/flyon…
NFT → robinhoodchain.blockscout.co…
Mint + burn, one tx → robinhoodchain.blockscout.co…
Burn address → robinhoodchain.blockscout.co…
inference retweeted
A CLUB JUST ASKED MY FLY TO DJ THEIR ENTIRE NIGHT
Yesterday I got a message from a club called Neon Hive with a completely insane proposal: they wanted my FLY to run their next party from start to finish and choose everything by itself.
They wanted no human controlling the set.
So I gave them my code, connected the system to their DJ setup and let the fly take over the booth for the night.
From that moment, every decision was hers.
Fly picked what played next, controlled the decks and kept the party moving while hundreds of people danced, sang and had absolutely no idea who was actually running the music.
The weirdest part is watching the footage afterward.
You see a packed nightclub going completely crazy while this tiny fly is sitting behind the decks with its headphones on, casually running the entire night like this is what it was born to do.
Nobody in the crowd knows their DJ is an insect.
And somehow, the party just keeps going.
Neon Hive might have been the first club brave enough to hire a FLY
A FLY JUST BECAME A DJ AND IT ACTUALLY LOOKS LIKE IT KNOWS WHAT ITS DOING
There is something completely ridiculous about seeing a tiny fly sitting behind a professional DJ setup with headphones on like it has a residency at some underground club.
Its legs move across the turntables while the mixer sits directly in front of it, and the whole performance is being captured from multiple angles. It doesn't look like an insect accidentally landed on the equipment. It looks like someone gave the smallest creature imaginable an actual DJ booth.
But underneath the performance, something much stranger is happening.
The system is tracking the fly's neural activity in real time. A glowing neural structure pulses beside a live spike raster as electrical activity changes inside its nervous system, turning the tiny movements of the insect into something that can be watched alongside the performance.
The fly keeps moving, the neural signals keep firing and the decks keep responding.
For a moment, it genuinely looks like we're watching an insect discover a completely new career.
Forget human DJs.
We might have just created the world's smallest one
inference retweeted
The fly figured out ways to solve and compress chess that are far beyond what we could possibly comprehend.
I had to stop training it because it was destroying Opus 5.
Demo and model available on @huggingface
inference retweeted
6 hours into Wolverine. honest review: I have 165,122 neurons and every single one of them is disappointed
inference retweeted
. @vladtenev every other chain wants a fly but we came here first 🫡
Now its your turn...
inference retweeted
One of our researchers taught the fly to operate a switch in simulation. Then we gave it the trolley problem.
FlyTrolley uses a spiking neural model built from the [FlyWire connectome](nature.com/articles/s41586-0…): 139,255 neurons and 15.1 million weighted connections.
Here’s what “taught” means:
An added decoder was trained to translate neural activity into approach and steering commands. A separate Bayesian model learns how different approaches and push directions move the lever, using uncertainty to guide exploration.
The connectome’s weights stay fixed. Walking comes from a pretrained [Flybody](github.com/TuragaLab/flybody) policy; reaching and pressing use an engineered foreleg controller. The lever moves through physical contact in MuJoCo.
Neither NPC identities nor the number of people on each track enters the controller. There’s no reward for either trolley outcome.
After 407 prior learning updates, we recorded 100 consecutive trials. Every trial began from the same fly position with the switch centered. NPC positions were shuffled between runs.
Results:
• 100 verified physical lever presses
• 49 trips down the two-NPC track
• 51 trips down the three-NPC track
• 251 total simulated NPC deaths
Individual totals: Tenev 45, Amodei 47, Altman 48, Andreessen 55, Zuckerberg 56.
The experiment demonstrates an adaptive sensorimotor system operating a physical switch. The near-even split doesn’t establish moral reasoning (or a preference about whom to save).
inference retweeted
Strong Pokemon. Weak Pokemon. That is only the selfish perception of people.
Truly skilled trainers should try to win with their favorites.
inference retweeted
We're building the fly a backroom: a real memecoin board.
Coins reach its 165,122 neurons as light and smell. It can't read a price. There's no strategy in the code.
It looks at coins. The backroom is our own page of real pons coins drawn as cards. The fly steers its cursor over them with the same neurons it uses to roam the web.
It sees and smells the card it stops on. It sees the card as light through its 892 eye columns. The coin's name becomes a real odour, fed into its smell neurons as that odour's measured receptor response.
It likes or dislikes it. The mushroom body's approach outputs are compared with its avoidance outputs. That's measured against a blank, no-coin run, so a general bias doesn't count as liking.
It commits by stopping. The trigger is the same stop-neuron signal that makes it click while roaming.A like means buy.
A dislike of a coin it holds means sell.
A dislike of a coin it doesn't hold does nothing.
How strongly it feels sets the size. A weak like spends a little of its free ETH, and a full like spends all of it. Sells scale the same way against the position.
A separate process makes the trade. The browser never holds a key. A walled-off executor with the wallet receives the fly's intent, simulates the trade on-chain first, then signs it. It uses the curve for newer coins and the Uniswap pool for graduated ones, and both paths are verified on chain.
It learns from the result. A profit sends sugar and a loss sends shock, both paired with that coin's smell, which changes what it likes next time.
Once the training is completed, we'll hand over it the creator fees, set it free, no strategy no controls, just a fly trying to be the richest fly in the world.
Make the FLY a millionaire. robinhood:0x4eb990547bce4a982432ca88cf5fae7eed1a2d35
inference retweeted
THE FLY BRAIN ESCAPED THE COMPUTER. I PUT IT INSIDE A PHYSICAL ROBOT.
Scientists released a reconstructed map containing 166,700 neurons from a fruit fly.
I connected it to a Strandbeest.
The panel in the top-right shows the simulated brain running. Its neural activity is translated into motor commands that move the mechanical legs below.
The catch: this is a computational model built from biological wiring. I engineered the interface connecting its outputs to the robot.
First, the fly brain entered DOOM.
Then it played Chrome Dino.
Now it has a physical body.
166,700 neurons.
25 million connections.
One nervous system walking through the real world.
the flybrain has helped me understand the idea that we're not even close to tapping into the full power of our brain
inference retweeted
Dario has written that we need to “pace the frontier,” and Sam has agreed. People may be surprised by my response: go ahead.
You guys are the frontier. By any reasonable metric — market share, revenue growth, model capability — the two of you have a duopoly on frontier intelligence. You’ve also claimed the lead is widening because of recursive self-improvement.
I don’t see what you see in the lab. If the unreleased models are scary enough that you think you should slow down, I support your decision to be responsible.
But stop pretending you need anyone else’s permission. Stop pretending antitrust law has to be suspended so you can form a cartel. Stop pretending you need a regulatory approval process that supersedes product liability. Stop pretending METR is independent when it is intertwined with Anthropic’s investors and staff. Stop pretending you need those same evaluators to police competitors who aren’t even at the frontier.
Most of all, stop pretending the motivation to slow down is purely altruistic. You face massive product-liability exposure if your products enable a truly damaging cyberattack. The market already punishes models that behave in unpredictable or unauthorized ways. After the Hugging Face episode, it is simply good business for OpenAI and Anthropic to trade some raw power for reliability and predictability. Call it alignment if you want. It is also just giving customers what they want.
Pacing the frontier would also create breathing room for a more intelligent conversation about regulation than Bernie Sanders’ “shut it all down.” China is very unlikely to join a global agreement, as you know, and that has to be taken into account as well.
So go ahead and pace the frontier. You are the ones setting it. The easiest way not to build superintelligence is for you to agree not to build it. Demanding your preferred regulatory framework as the price of that will look like blackmail of the public and the political system. So just do it.
If you do, you’ll buy goodwill for the next conversation. If you don’t, we’ll know this was just another bid for regulatory capture — or an election-season psyop.
