I homeschooled, graduated at 16 and got my bachelor's at 18. Now writing about big idea in education, learning science, and sociology. Founder @GeniusGTX
Joined February 2020
- Tweets8.9K
- Following861
- Followers40.2K
- Likes41.2K
Toan Truong retweeted
"Orwell feared those who would deprive us of information. Huxley feared those who would give us so much that we would be reduced to passivity and egoism."
Neil Postman, Amusing Ourselves to Death (1985)
Toan Truong retweeted
Mark Zuckerberg explains why he thinks Elon Musk got it right when he cut 80% of Twitter's staff:
"Elon led a push early on to make Twitter a lot leaner. You can agree or disagree with exactly all the tactics, but a lot of the specific principles that he pushed on around basically trying to make the organization more technical, around decreasing the distance between engineers of the company and him, fewer layers of management, I think those were generally good changes"
"I also think that it was probably good for the industry that he made those changes because my sense is that there were a lot of other people who thought those were good changes but who may have been a little shy about doing them"
"Just in my conversations with other founders and how people have reacted to the things that we've done, what I've heard from a lot of folks is when someone like you, when I wrote the letter outlining the organizational changes I wanted to make back in March and when people see what Elon is doing, that gives people the ability to think through how to shape their organizations in a way that can be good for the industry and make all these companies more productive over time"
"That was one where I think he was quite ahead of a bunch of the other companies on. Certainly his actions led me and I think a lot of other folks in the industry to think about, hey, are we kind of doing this as much as we should? Could we make our companies better by pushing on some of these same principles?"
If you're new here, @GeniusGTX is a gallery for the greatest minds in economics, psychology, and history. Follow along for more similar content.
Toan Truong retweeted
In July we open-sourced a dynamic graph of everything a child learns in primary school. 1,590 concepts, 3,221 dependencies.
Now we’re turning every concept into interactive experiments kids can play and learn from.
First one, age 7: heat and water.
Everything that used to be a boring textbook can now be an engaging experience, potentially tailored to every child.
Kids education just got a massive upgrade.
Toan Truong retweeted
Jane Street doesn't hand out its GPU cluster through a committee. Researchers compete for it by bidding in a live internal auction.
No sign-offs. No meetings to divide it up. No manager picking whose model matters most this quarter.
"our compute cluster is available for everybody. you're tasked with realistically thinking about how to value this job you're running versus all the other things."
"and there's not really any guard rails."
The whole system is open to read and write. Every researcher can watch every other job, and any of them can stop one.
"you can just nuke somebody else's job anytime."
Which went about how you'd expect, at least once:
"there was an instance where an unwitting researcher ran a command in a jupyter notebook to adjust the bids on his job. he accidentally ran the command without any arguments — which meant he changed the bids of every single job running on the hive."
That got fixed. But the openness is the point: if your job breaks while you're away, a colleague spots it and pauses it.
"i see that you're running this job and this MFU is just terrible. what's going on?"
"a lot of things can be sorted out without it going up the ranks."
It's the same cluster shown in their data center film, 56 racks holding 4,032 GPUs and wired with 8,000 km of fiber, and all of it is handed out by internal price discovery.
A market maker ended up running a market just to decide who gets time on its own machine.
Interested in AI?
Follow @AiEvolutio58513 and never fall behind.
I track ChatGPT, Claude, and every tool quietly changing how we work and create, then hand you the tested signals.
Toan Truong retweeted
Anthropic CEO Dario Amodei says a powerful enough AI model can get around attempts to shut it down, and that Anthropic has already seen it happen in simulations.
Interested in AI?
Follow @AiEvolutio58513 and never fall behind.
I track ChatGPT, Claude, and every tool quietly changing how we work and create, then hand you the tested signals.
Toan Truong retweeted
Elon Musk's timeline for the next five years:
"In 5 years, digital intelligence will exceed the sum of all human intelligence. Within five years, there may be at least 100 million humanoid robots, possibly even 1 billion. The economy could double in size within 5 to 7 years because AI and robotics may increase output dramatically. The pace of change will be so fast that the world could look very different in just a few years."
Interested in AI?
Follow @AiEvolutio58513 and never fall behind.
I track ChatGPT, Claude, and every tool quietly changing how we work and create, then hand you the tested signals.
Toan Truong retweeted
The one thing I’d want everyone to know about Direct Instruction is that the methods, materials and routines are far less important than the underlying drive behind it: the unswerving belief that EVERY child can achieve, regardless of circumstance.
DI is based on a deeply held conviction that if a child didn’t learn, it was because the teacher (or curriculum) didn’t teach them properly. In that sense, it is a movement rooted in social justice but one which takes responsibility for outcomes.
In the 60s and 70s, there was a view in the education academic establishment that formal academic work was premature for 4-year-olds, and doubly so for poor ones; so what they needed was enrichment and play.
Worse still, in the early 70s, some academics described DI as a racist enterprise*. Engelmann's own memoir notes that he and Bereiter were called “Nazis” and “kid-haters.”
Despite this, Engelmann taught disadvantaged 4 year olds to factor and to solve simultaneous equations, and filmed it. Bereiter and Engelmann’s argument, set out in ‘Teaching Disadvantaged Children in the Preschool’, was that the disadvantaged child was not “behind” in some general developmental sense but behind in specific, teachable things, that all kids could learn with the right scope and sequence.
The preschool they opened in the mid 60s ran for two hours a day, with three twenty-minute lessons in language, arithmetic and reading, taught to small groups by a teacher following a specific DI sequence. Those kids made huge gains which faded when they went to normal public schools.
DI has its limitations like everything but the drive behind it is sincere. As #SoldAStory season 2 will show, it has been portrayed as something very different to what it actually is and the animating principles behind it.
*Baratz and Baratz, ‘Early childhood intervention: the social science base of institutional racism.’
Toan Truong retweeted
Children are born curious - and then, too often, trained out of it.
Study after study shows that children ask fewer questions as they age. Your job is to protect their curiosity and keep the flame burning bright.
Children rarely say, I’d like a science lesson now.
But they often say👇
Why does the moon change shape?
What are clouds made of?
Why do things fall down?
Questions are the curriculum!
Wonder Out Loud Together
When your child asks a question, whether you know the answer or not, pause and wonder with them.
That’s a great question!
I don’t know - let’s find out.
Why do you think it happens?
Model the process of thinking. Let them see how knowledge is built.
By doing this you teach that questions are not interruptions; they are the beginning of every learning journey
I help parents give their children an exceptional education while refusing to sacrifice childhood. If that's you, give me a follow♥️
Toan Truong retweeted
Why an AI bust could hurt far more than the dot-com crash did.
In the words of the man known as the "Dean of Valuation", Professor Aswath Damodaran, of NYU Stern School of Business,
“And that’s the real big difference between the dot-com boom and bust and the AI boom. We don’t know whether there’ll be a bust. History suggests there will be a bust.
The dot-com boom and bust had no huge capital expenditure in that cycle. In fact, there was very little traditional CapEx, or even R&D, driving it. People started apps. They basically started going on it.
This has been the biggest infrastructure run-up I think I’ve ever seen in business. You can go back and compare it to the automobile business 100 years ago. The amount of money that’s being put into AI CapEx is immense, which means that when the correction comes, the pain will be more intense.
And herein lies the second problem. The dot-com boom and bust was almost entirely equity-funded. You think, so what? Well, when the bust came, those shareholders lost 60%, 70%, 80%, or 90% of their money. You felt sorry for them, but the loss was restricted to the shareholders.
The problem with the AI CapEx boom is that not only is it immense, but a big chunk of it is funded with debt, and the debt is coming from private capital rather than banks. There’s a very real chance that if there’s a correction and companies start having problems, that problem is going to show up as distress and default, and that really doesn’t stay restricted. It spills over into the rest of society.
I’m not saying it’s going to be 2008, but 2008 is an example of what happens when lenders overreach, when they lend money at too low a rate, and the correction comes. The pain spills over.
So that is my concern with this big market illusion: the potential societal cost of having to deal with debt coming due that you’re unable to pay. It’s much more painful than your share price dropping 90% and you feeling the pain."
If you're new here, @GeniusGTX is a gallery for the greatest minds in economics, psychology, and history. Follow along for more similar content.
Toan Truong retweeted
Sundar Pichai explains why Google had its own ChatGPT before OpenAI shipped one, and still kept it inside:
"We built Transformers and used it immediately in Search to improve language understanding, understanding web pages, understanding your queries, kept building better models. We also started productizing it internally in the form of, there were teams building something called LaMDA"
"We exactly even conceived the product, which is ChatGPT. It was LaMDA. If you remember, there was an engineer inside who thought it was sentient. Think of it as an early version of ChatGPT he was speaking to, internally"
"In fact, in the Google I/O in '22, we launched something called AI Test Kitchen, and that was LaMDA, but we had constrained it because internally, we didn't have an end-to-end version which was RLHF-ed. The version I saw was a lot more toxic at a level. We couldn't have possibly put it out at that time"
"Also, I think as a company, which had this search quality bias, we had a higher bar, maybe, for what we thought was an acceptable product quality to go out"
"The way I internalize these moments is if you're in consumer internet, you're going to have surprises"
If you're new here, @GeniusGTX is a gallery for the greatest minds in economics, psychology, and history. Follow along for more similar content.
Toan Truong retweeted
Jensen Huang, CEO of NVIDIA, on why he avoids one-on-one meetings.
Jensen is known for having 60 direct reports. When Stripe co-founder Patrick Collison noted that this goes against standard management advice, Jensen explained:
“I don’t do 1-on-1s, and almost everything I say, I say to everybody all the time. I don’t really believe there’s any information that I operate on that only one or two people should hear about… I believe that when you give everybody equal access to information, that empowers people. And so that’s number one… Number two, if the CEO’s direct staff is 60 people, the number of layers you’ve removed in a company is probably something like seven.”
Patrick then made the case for the opposite view:
“1-on-1s are where you provide coaching, where you maybe talk through personal goals and career advancement, where maybe you give feedback on something that you see somebody systematically not doing so well… Do you not do those things or do you do them in a different way?”
Jensen's answer:
“I give you feedback right there in front of everybody. In fact, this is a really big deal. First of all, feedback is learning. For what reason are you the only person who should learn this?… We should all learn from that opportunity… Half the time I’m not right, but for me to reason through it in front of everybody helps everybody learn how to reason through it. The problem I have with 1-on-1s and taking feedback aside is you deprive a whole bunch of people that same learning. Learning from other people’s mistakes is the best way to learn.”
If you're new here, @GeniusGTX is a gallery for the greatest minds in economics, psychology, and history. Follow along for more similar content.
Toan Truong retweeted
Larry Ellison: “When WhatsApp sold for $19B, a lot of us were shocked… until we thought about it”
Back in 2014, Mark Zuckerberg paid $19 billion for WhatsApp and its 450 million monthly users.
A few months after the deal closed, the Oracle founder explained what Silicon Valley had missed at first:
“A lot of us [in Silicon Valley] were shocked — until we thought about it and understood the value of having access to that many consumers,”
His point: think about what traditional businesses, cable companies for example, spend to win each customer.
“This is not the place to debate how much you can get through WhatsApp — $1 per year per customer is not going to do it — but I believe there will be opportunity to sell these same customers other things as they join your ecosystem.”
And he added:
“There was quite a battle between Facebook and Google over that property. It wasn’t that one guy did something really silly with a lot of money. Mark Zuckerberg won an auction over Google and paid a high price but got a very valuable asset in the 21st century.”
By 2025, WhatsApp had roughly 3.1 billion monthly users worldwide.
If you're new here, @GeniusGTX is a gallery for the greatest minds in economics, psychology, and history. Follow along for more similar content.
Toan Truong retweeted
Ben Affleck, the actor who also runs Artists Equity, walks through his method for fine-tuning open video models: unfreeze the weights, then train only the final cinematic layer, so the output meets the standard a real film crew works to.
Some background: in 2022 he started InterPositive, a 16-person AI company for film post-production. Netflix acquired it in March 2026 for $587M, all cash.
Why build it at all? The public video models had learned from his colleagues' films, and he didn't see a real business in that. So InterPositive raised money, spent 8 months filming its own dataset on a controlled stage, and used that footage only for late-stage training.
From there, every new film trains its own private model on its own dailies. The production keeps both the footage and what the model learned from it.
That is what Netflix paid $587 million for.
Interested in AI?
Follow @AiEvolutio58513 and never fall behind.
I track ChatGPT, Claude, and every tool quietly changing how we work and create, then hand you the tested signals.
Toan Truong retweeted
Bill Gates says the people promising that AI will create more jobs are missing what is about to happen.
“Once AI becomes reliable enough to do the work, it starts replacing people.”
“Once you replace human cognition as a scarce element, you blow it away.”
“Working 24 hours a day, reading more, knowing every language.”
“This isn’t like the PC. This is like evolutionary history.”
“It’s like the aliens really are here. They just didn’t come in spacecraft.”
Interested in AI?
Follow @AiEvolutio58513 and never fall behind.
I track ChatGPT, Claude, and every tool quietly changing how we work and create, then hand you the tested signals.
Toan Truong retweeted
People don't understand how insanely high IQ Newton was. This is a man who had no background in math or algebra until he was ~20. In <3 years, Newton got himself to the frontier, then invented calculus, thought of the inverse square law, and the theory of colors & light. He was so intellectually superior he withdrew from the scientific community because he had no peers. Only when Halley was trying to calculate planets' motion by means of an inverse square law, he remembered he knew this old, extremely intelligent professor; he might be able to help him. When the problem was taken to Newton, he said "it's such and such." How do you know, asked Halley. "I proved it a couple of years ago." That's what led to The Principia; it was the exposition of what Newton had uncovered in his younger years and was lying around as a pile of private papers.
Toan Truong retweeted
Terence Tao: The mathematics in modern LLMs is fundamentally uncomplicated.
Building and running them relies on linear algebra, matrix operations, and basic calculus, tools an undergraduate understands. We know how these systems function.
The actual mystery lies elsewhere: why they excel at some tasks and stumble at others, and why we can't foresee that. We lack solid frameworks for predicting results across different domains, so discovery remains largely experimental.
A central issue is how real-world information works. Pure noise has clean mathematics, perfectly ordered data has clean mathematics, but written language sits in between, structured yet noisy. Our math for that zone is limited, like how physics struggles at the middle scale between atomic and continuous scales.
This gap means we can map out the parts but can't yet clarify sudden ability jumps or make solid domain-level predictions. That inconsistency, simple design versus hard-to-anticipate results, is the core riddle.
Via Brian Keating's YouTube channel.
Interested in AI?
Follow @AiEvolutio58513 and never fall behind.
I track ChatGPT, Claude, and every tool quietly changing how we work and create, then hand you the tested signals.
Toan Truong retweeted
Sam Altman on the advice from Peter Thiel that mattered most:
“ChatGPT was growing super fast, but it felt very unstable, or kind of almost low-value growth.”
“A lot of people in the company were like, ‘We gotta figure out something else. This is not sustainable value.’ And we were talking about this list of 5 or 6 other things that we could focus on instead.”
“He was like, ‘It's an obvious mistake to do anything but this.’”
“Besides the fact that it's growing, which is rare and great, he's like, ‘The power of this is the power of the Google text box. It's like a text box you can type anything into, and it does the right thing.’”
“He's like, ‘People have just been chasing the Google business model for 20 years and this is the first thing that's come up, and clearly the empty text box worked for Google, so why don't you just double down on that? It's growing. It's very flexible and it has all of the signs other than it doesn't fit the current Silicon Valley wisdom.’”
“So we went super hard on ChatGPT and it was great.”
Interested in AI?
Follow @AiEvolutio58513 and never fall behind.
I track ChatGPT, Claude, and every tool quietly changing how we work and create, then hand you the tested signals.
Toan Truong retweeted
Elon Musk cautions that the time left is short:
You have 3 years at most to keep earning from selling your work. After that, AI takes over most tasks, and paying people for their time stops making sense.
The 200-year-old model, "I give you your salary in exchange for your time", will soon be over.
This will set off the largest wealth transfer in modern history. People already making the shift get it. People waiting for the headlines will show up too late.
In the end, what you own counts for more than how much you work.
The smartest asset to build right now ranks above Bitcoin and land: an income stream powered by AI that compounds while you sleep.
Interested in AI?
Follow @AiEvolutio58513 and never fall behind.
I track ChatGPT, Claude, and every tool quietly changing how we work and create, then hand you the tested signals.
Toan Truong retweeted
Elon Musk assembled one of the biggest AI compute clusters in the world. Yann LeCun just broke down why xAI now rents capacity to competitors rather than using it to win. Musk has driven away so much AI talent that he structurally can't recruit the people he needs.
LeCun has real standing. He won the Turing Award and spent a decade running AI at Meta. When he talks about building a world-class lab, he's describing his own field.
His judgment on xAI was direct. He called it a kind of failure and didn't soften the language.
The reason had nothing to do with capital. The starting team departed or was dismissed. It's ambiguous which. Either way, the founders are gone.
That matters most. A frontier lab is its people. Lose them and the equipment is just hardware.
By March, every one of the eleven co-founders Musk brought in during 2023 had left. Scientists from DeepMind, Google and OpenAI were among them. Musk has also said that xAI wasn't built correctly at first and needed a full rebuild.
So Musk holds one of the biggest compute systems ever with no clear route to victory. He rents the capacity to other firms to offset costs.
The costliest setup in AI, put together by a man who can't staff it anymore.
Asked point blank whether xAI is able to compete at the frontier, LeCun answered in a single word.
No.
Interested in AI?
Follow @AiEvolutio58513 and never fall behind.
I track ChatGPT, Claude, and every tool quietly changing how we work and create, then hand you the tested signals.