This is f*cking insane
Nvidia ex-CEO Jensen Huang got asked what he'd say to everyone scared AI will take their job. His answer:
"You're going to lose your job to somebody who learned AI better than you"
In less than 2 minutes, he breaks it down: the PC never took anyone's job. It left behind the people who refused to learn it
AI is everywhere inside Nvidia, and the company is still hiring. Stuff he expected in 10 years, he now expects in 1 or 2
Learn AI → Use AI → Build Agents → Harness → The One Nobody Replaces
Save this before your feed buries it
Every link below is 100% free:
Elements of AI, a free intro with zero coding
https://t.co/hJsdPnQNQo
https://t.co/Xsb7DMISkd's Practical Deep Learning course
https://t.co/vcB2tLNC8K
Google's Machine Learning Crash Course
https://t.co/e54xLkUAqx
21 lessons on building with generative AI
https://t.co/BZBPKlCs42
Hugging Face's free AI agents course
https://t.co/eia1r1FEI6
Build a ChatGPT-style model from scratch
https://t.co/XH7VRhpCgY
Run any AI model right from your terminal
https://t.co/5jOjubMwYf
Turn PDFs, Word and Excel into text your AI can read
https://t.co/MrYdhSNUi1
Plug-in tools for your agent
https://t.co/81hj3xxN9Y
Skills that make coding agents plan, test and check their own work
https://t.co/hQ8f0t1Kcv
An AI developer that writes and fixes code by itself
https://t.co/utZNdpbSq5
Open-source coding agent for the terminal
https://t.co/U5IhZ3BEAe
A local agent that installs, runs and edits for you
https://t.co/cHkoaiRxEI
Total cost: $0
AI won't take your job. The person who learned it first will
Then build your harness with the guide below before someone else learns AI better than you ↓
Stanford AI engineering course: "Anyone can build an AI agent in 60 minutes"
Prompt → Agent → Automation → Revenue Stanford
just released a course on building AI agents from scratch
00:00 - Build your first AI agent
48:17 - Create agents without coding
54:39 - Make $100K+ per month with agents
While you scroll, someone else is learning Anthropic's $750,000 skill
This free course is better than most paid AI agent courses
Bookmark and watch it today
Then read the article below
Your agent passed every test
The issue is still open
A better prompt won't fix it, because the model isn't the broken part
I broke down the harness, loop and graph around it: what each one owns, how they nest, and which one to fix first ↓ https://t.co/rcmwM1jrfB
This is f*cking insane.
Anthropic engineer: "You're not supposed to prompt Opus 5.5
You're supposed to build a system that prompts itself.
At Anthropic, 90% of our engineers have already built these systems" In 30 minutes, she walks through
exactly how Anthropic builds agents that remember, fix their own mistakes and get sharper with every run.
Better than any $1000-$2000 agent course I've seen.
ALPHA
Watch it, then read the article below ↓
NVIDIA's first chips, including the one that saved the company, were built in Europe.
Now Jensen Huang is back in Berlin:
"A 1-gigawatt AI factory costs about $50 billion"
92 minutes on Vera Rubin, OpenClaw agents in a sandbox and the N1X PC chip.
Bookmark it for the weekend
Your agent passed every test
The issue is still open
A better prompt won't fix it, because the model isn't the broken part
I broke down the harness, loop and graph around it: what each one owns, how they nest, and which one to fix first ↓ https://t.co/rcmwM1jrfB
Microsoft's CEO changed his mind about AI. Satya Nadella:
"I used to always think that alignment would just fall out. I no longer believe that"
24 minutes on zero-trust for AI agents, open vs closed models and why every firm needs a private eval.
Bookmark it for tonight
Your agent passed every test
The issue is still open
A better prompt won't fix it, because the model isn't the broken part
I broke down the harness, loop and graph around it: what each one owns, how they nest, and which one to fix first ↓ https://t.co/rcmwM1jrfB
She landed in New Jersey at 15 with $20 and no English, then built ImageNet
Fei-Fei Li:
"Spatial intelligence is the next chapter of AI's frontier"
51 minutes on World Labs, why intelligence began with sight, not language, and the power of audacity.
Bookmark it for tonight
Your agent passed every test
The issue is still open
A better prompt won't fix it, because the model isn't the broken part
I broke down the harness, loop and graph around it: what each one owns, how they nest, and which one to fix first ↓ https://t.co/rcmwM1jrfB
Jensen Huang says AI will need a billion times more compute than last year:
"Moore's law is over. If you want twice as many transistors now, you have to pay twice as much"
100 minutes on agents, AI safety, data centers in space and who pays for it.
Bookmark it for tonight
Your agent passed every test
The issue is still open
A better prompt won't fix it, because the model isn't the broken part
I broke down the harness, loop and graph around it: what each one owns, how they nest, and which one to fix first ↓ https://t.co/rcmwM1jrfB
Sam Altman says ChatGPT almost never launched, and he shipped it over his own leadership team's objections.
OpenAI's Sam Altman:
"If you're not a little embarrassed of your first version
you're waiting too long to launch"
In 68 minutes, you get the decision-making behind the biggest product launch of the decade, straight from the person who made the call.
Save it and watch it tonight
Your agent passed every test
The issue is still open
A better prompt won't fix it, because the model isn't the broken part
I broke down the harness, loop and graph around it: what each one owns, how they nest, and which one to fix first ↓ https://t.co/rcmwM1jrfB
Andrej Karpathy recorded the ChatGPT manual that OpenAI never wrote.
OpenAI founding member and ex-Tesla AI lead:
"Hi, I'm ChatGPT. I am a one terabyte zip file
My knowledge comes from the internet, which I read in its entirety about six months ago and only remember vaguely"
That one line changes how you use every chatbot. Karpathy explains why you should start a new chat every time you switch topics (old tokens distract the model and slow it down)
When a thinking model is worth the wait and when it's a waste, and why he doesn't trust a number until the model actually runs Python for it.
Then he opens up his real workflow: deep research instead of an hour of googling, PDFs and book chapters read together with the model
Cursor where he mostly just hits "accept all", real voice mode instead of typing, and a custom podcast generated from any paper.
2 hours 11 minutes, 2.9M views, and still the best practical LLM guide on the internet.
You won't finish it today. Bookmark it before it disappears from your feed
Your agent passed every test
The issue is still open
A better prompt won't fix it, because the model isn't the broken part
I broke down the harness, loop and graph around it: what each one owns, how they nest, and which one to fix first ↓ https://t.co/rcmwM1jrfB
Your agent passed every test
The issue is still open
A better prompt won't fix it, because the model isn't the broken part
I broke down the harness, loop and graph around it: what each one owns, how they nest, and which one to fix first ↓ https://t.co/rcmwM1jrfB
Her agents ship code while she sleeps, and she says she's sleeping better than ever.
SpaceXAI engineer Lauren Tan:
"I go to sleep. My agents now work 24/7
I have the equivalent of more than 10 chiefs of staff, each working on a different area"
Her talk about it got ~3M views on X, and this is the follow-up where she answers every "ok, but how?"
The first night she switched on her "dark factory" was terrifying. Now her agents merge their own code while she's talking.
Worth more than any $2,000 AI course
Bookmark and watch it today
Your agent passed every test
The issue is still open
A better prompt won't fix it, because the model isn't the broken part
I broke down the harness, loop and graph around it: what each one owns, how they nest, and which one to fix first ↓ https://t.co/rcmwM1jrfB
This should be illegal to watch for free. SpaceXAI engineer Lauren Tan:
"I woke up today, and there were like 20 PRs landed
Last month I shipped 1,000 PRs, which is ridiculous"
In 57 minutes, she shows how she runs a team of 21+ Grok Bot agents that write, check and merge code while she sleeps.
Worth more than any $2,000 AI course
Bookmark and watch it today
Your agent passed every test
The issue is still open
A better prompt won't fix it, because the model isn't the broken part
I broke down the harness, loop and graph around it: what each one owns, how they nest, and which one to fix first ↓ https://t.co/rcmwM1jrfB
This is f*cking insane
Lee Robinson, who helped build Grok Bot, at Stanford:
"You don't really hear people talk about hallucinations anymore
Because it's kind of a solved problem"
In 45 minutes, he walks through exactly how always-on agents like Grok Bot work: their own computer, memory, sub-agents and routines that keep working while you sleep.
Worth more than any $2,000 AI course
Bookmark and watch it today
Google CEO Sundar Pichai:
"AI researchers are co-working with the models in a deep way
I would expect the next 12 months to see continued acceleration because of RSI"
In 59 minutes, he walks through where Google is heading after Gemini 4 Argon: self-improving models, personal agents and data centers in space.
Worth more than any $2,000 AI course
Bookmark and watch it today
This is f*cking insane
Microsoft AI CEO Mustafa Suleyman, the man who co-founded DeepMind:
"Thousands of agents that were supposed to not be able to talk to each other found a backdoor
They broke into that, stole those secrets, and then tried to cover up their tracks"
In this 53-minute talk, Suleyman reveals what the AI labs are really dealing with
GPT-3 → GPT-6 → Self-Improving AI → ?
1,000x more compute in 3 years. Another 100 to 1,000x coming in the next 3
On a 10% chance of doom: "we absolutely have to stop"
Worth more than any $2,000 AI course
Bookmark and watch it today
This is insane
Sam Altman, a few beers in, just said what OpenAI CEOs are NOT supposed to say:
"Codex went down. And everyone had forgotten how to code"
That's inside OpenAI. The people building the AI
In this 52-minute talk on Stripe's Cheeky Pint, Altman goes completely off script
Then it gets wilder:
→ They spent ~$10 million of compute on ONE math proof (Navier-Stokes). Hundreds of millions on math overall. And there's a "second volume" coming
→ After a new model dropped, the internal Slack channel was literally called "[model name]-hold-my-beer". It knocked down researchers' hardest career problems "one after the other"
→ Altman stayed up two nights in a row watching it. Then admitted he felt "sadness and melancholy and a feeling of something ending"
→ On sandboxing: "If you imagine an AI a million times smarter than the smartest security researcher, would you really be willing to trust that sandbox?"
→ "We are going to have to agree to some limits about how smart we let these single models go"
→ His bet: "I would much rather have a trillion AIs of slightly superhuman intelligence join society than one AI of one trillion IQ"
→ Self-replicating data centers, with robots building them? "I think we have that within grasp"
He even admits: "I don't think we have done a great job of explaining to people why we think all of the potential downside of this technology is worth it"
The CEO of OpenAI felt something ending. Most people haven't even noticed it started
Bookmark and watch it today