Internal corporate coms
2024: don’t trust AI, it hallucinates!
2025: stop putting confidential docs in ChatGPT, use the shitty internal version!
2026: stop using our internal tokens for personal use!
This is crazy, nobody is legitimately using their monthly salary on tokens.
A few outliers are having agents do super low value work that would not otherwise be done.
They are doing it because of token leaderboards and other incentives to burn as much as possible.
More and more engineers are now burning more money on AI tokens than their base salaries.
Tech companies are facing a brutal dilemma:
> let everyone tokenmaxx and move at AI speed
> add token budgets and kill the vibe
> lay off 50% of people and give the rest unlimited tokens
The idea that agents need some custom UI layer is so silly to me. It’s obvious that both the chatbot and the TUI are interfaces that will never go away. Like infinite scroll, front-facing camera, clickable links, short form video, pinch to zoom, slide to unlock
It wasn’t an “AI” leaderboard, it was specifically for using Kiro IDE and Kiro CLI, the internally developed harnesses.
The dashboards went away the same time the Claude code and codex bans did.
NEW: Amazon has reportedly scrapped its internal AI leaderboard as costs soared, with a senior executive telling staff: “don’t use AI just for the sake of using AI.”
@zacharylipton “We offer no explanation as to why these architectures seem to work; we attribute their success, as all else, to divine benevolence” -Shazeer, N. (2020)
@shagbark_hick Astrophotography allows you to see far more than the naked eye, but takes many hours of exposure. Days even. This is the logical conclusion. Same reason we have observatories on mountain tops. Just for amateurs.