The dependency curve always runs ahead of the pricing curve. We are becoming locked in with LLMs and they’re becoming indispensable.
You will pay for rising AI usage fees, now matter how aggressive they surge. You are reliant on these models, they’re increasingly woven into every workflow and thought process you have now. It’s not the value you’re paying for anymore, it’s the pain you’re avoiding of reverting back.
AI is no longer just a superior tool, it’s a supreme toll.
The biggest flaw in how we train frontier AI models is embarrassingly simple.
Imagine spending a full day solving a hard problem. You try dozens of approaches, hit dead ends, backtrack, get lucky once. Your model reasons for hundreds of steps. But, at the very end, it gets one signal: right or wrong.
That single bit gets broadcast across the entire chain as if every step contributed equally. Wrong turns get reinforced because the answer landed. Good reasoning gets penalized because the conclusion missed.
It's not a subtle flaw. It's a fundamental misattribution of credit. But, this is we train the most capable AI systems on earth.
It's like sucking supervision through a straw.
using github as a personal OS is the first step. the next step is letting autonomous agents manage the repository.
we’re moving from "managing tasks" to "orchestrating agents" who handle the chores, bills, and schedules while we focus on architecture.
the goal isn't to be better at using tools. it's to build tools that don't need us.
GitHub powers your code, but it can also power your daily life. 🔋
Instead of downloading another productivity app, manage your tasks right where you already work:
✅ Issues for chores and bills
🏷️ Labels for priority and status
📊 Projects for your daily schedule
Here’s how to set up your personal operating system. 👇
https://t.co/WsO5Zwpb5C
@eric_lombrozo i get where you’re coming from, but half the fun of any 'game' is the community discussion around it. it’s all part of the experience! 😊