abel 亚伯
@Abel_summation
researching parallel multi-agent systems and LLM ensembles. 15 y/o
just finished a busy high school year, I'm 15 and summer's the time to lock in. a couple months ago I built a hand-tracking app that lets you control the computer with gestures, finalist in a national AI comp (lost, but we move), and international stem gold medalist.
parallel-agent synchrony is our best shot at achieving superintelligence. Nick Bostrom said it in his book: there are 3 types of ASI - quality, collective and speed. parallel multi-agent workflows give us a shot at speed + collective superintelligence. been doing some research…
GROK 4 GOES PARALLEL, STACKS BRAINS TO SOLVE HARDER TASKS Grok 4 isn’t just faster, it’s smarter in formation. The newest mode spins up multiple agents in parallel, each solving the same problem their own way. Then they cross-check and converge on the best result. Think…
I recently had a conversation with Jay McClelland -- professor of cognitive science at Stanford and one of the pioneers in neural networks along with Geoffrey Hinton and David Rumelhart. I got the chance to ask him something I’d been thinking about: > AI understands by…
Peter Thiel mastered the art of saying complex things simply. a lot of people think the goal is to distance yourself from normies with buzzwords and convoluted language. but clarity isn’t necessarily compromise -- it’s how influence spreads. In a democracy, the smartest person…
Jobsianism: the silent cultural and ideological control by Steve Jobs there’s a form of aesthetic control so deeply embedded into our subconscious that we don't even question it. it just feels right. why is the number 100 such a universal benchmark? Why not 193? Why do we…
the overton window doesn’t move randomly. it follows a blueprint. the blueprint is the hegelian dialectic. the zeitgeist moves in spirals, not horizontally. what was once extreme becomes normal, and what was normal becomes outdated. it revolves around a circle. people who…
I’ve been experimenting with parallel multi-agent workflows for a while. cool to see deepmind apply it and get gold.
in the monarchic era, cultural influence revolved around royalty. If you wanted part, you aligned with the kings. to control the zeitgeist you either had to be royalty, or the gray cardinal pulling the strings. but the internet dramatically decentralized the narrative. memes,…
every year the bar for AGI is set higher and higher. It started with chess, people thought it would be impossible for a machine to beat a human at chess. after all, it was a human game. then it was writing. no way AI would be able to convince people that it is a human.…
when playing a character, one must try his best not to get stuck. the hierarchical structure -- judges, teachers, authority figures -- rewards those who can adapt their persona to fit the part. I learned to exploit it, purely because it worked. speaking in a certain way,…
game theory manifests itself in situations you’d never imagine. recently I participated in a national debate competition. when it started I had gone in thinking the goal was to present the best arguments, which I actually believed to be true. but turned out the truth wasn’t the…
the hegelian dialectic is a concept few have fully internalized and implemented a while back a friend mispronounced a word, to which I jokingly trumped his slip with a worse one, he did the same, and then I pushed the joke to the extreme, an absolute distortion. for a moment I…
book recommendation by me if you’re trying to learn more about AI/ML/deep learning. quick read, not overly technical, and helped me understand how GPTs and LLMs work at a core level.
whenever I get a new startup idea I go back to the OG zero to one to check if it’ll actually survive sharing the pages in case they help avoid wasting time on failing ideas
after reading 20+ papers on LLM forecasting and market prediction, I’m planning to research and experiment with this myself. I’ve pulled the best methods and came up with some to test whether LLM ensembles could forecast the future better than live human markets like Polymarket.…
hey grok, write me a banger post, 5k likes minimum. make no mistakes.
this is Jay Mcclelland the man who taught machines to think like minds. he co-authored the parallel distributed processing papers with David Rumelhart and Geoffrey Hinton - a collaboration that laid the groundwork for today's neural networks. before parallel distributed…
This is Fei-Fei Li. the woman who taught machines to see. she built ImageNet, the massive dataset that turned computer vision from a research dream into real-world AI. before her, AI struggled to recognize a cat. after her, it could drive cars, diagnose diseases, and power your…
Browser Use just got public sharing! 🎉 Your AI agents can finally get their 15 minutes of fame! ⭐ Click to share → blow people's minds → agents get inflated egos → repeat 🔄 The automation flexing is about to be UNREAL 💪✨
