🔥The 9th Round of Easy Loan, Earn $40 Reward is in progress❗️ ⏰ Promotion Period: January 15th - Feburary 15th, 2025 👉 Register now and check more details at gate.io/campaigns/358

psa: you have the permission to dream

my co-founder Paul Eremenko talked w/ Sonya from Sequoia Capital team about what we're building at p-1 ai great listen!

the real scaling laws usa needs to work on a better coefficient new breakthroughs are needed

is this normal?

shipping it P-1 AI

if your product version is >= 1.0 before 500M ARR you're ngmi

more is different

we thought we'd get paperclip maximizers, but we ended up getting sycophants

we're now basically building infinite RL environments - a simulation of our world, and once that grand project is completed, we'll have given birth to a superintelligence at that same point in time, a new universe is born, and evolution will start in this new world, ultimately

getting to human-level performance took: 15 years on MNIST 7 years on ImageNet 9 months on GLUE and now it feels we're talking days/weeks (and benchmarks are increasingly more challenging) once it becomes real-time, do we even care what you call it?

twitter algo changing is the closest analog to how life would have to adapt if the laws of physics suddenly changed

Planning is the cousin of procrastination

There are two kinds of people in this world: those who count 1, 2, 3 - and those who count 1, 3, 4

🫡

This is a really great point and something I totally forgot about. I've been often thinking there may be future optimization algorithms that may deviate from SGD or be something of a hybrid approach. However if you keep stacking parameters local minima are less of an enemy