Last week my tram was told that they are monitoring our AI usage and those that are not using it at are risk. We were shown a dashboard that tracks general ChatGPT usage and Codex usage.
This week an email comes out instructing us on the proper use of AI, which model to select based on usage cost.
So wait, you want us to use it but in the same breath you are worried about the usage cost?
Id--ts
I do love me some good French toast though!
define what wastefully abusing means
It means you’re using it wrong. This guy knows what’s up:
https://www.bloomberg.com/news/articles/2026-06-10/apollo-s-zito-says-too-much-ai-spending-is-for-low-iq-tasks
Apollo’s Zito Says Too Much AI Spending Is for ‘Low IQ’ Tasks
Amanda CantrellJune 10, 2026 at 11:30 AM EDT
Apollo Global Management’s John Zito said too many companies are deploying artificial intelligence tools for relatively mundane tasks that don’t justify the significant costs and computing power.
“Our IQs are so low that we’re actually using” AI tools “to check out the recipe for, you know, French toast,” said Zito, co-president of Apollo Asset Management. “That’s where you’re seeing the prices go up.”
…
Also, lets not forget that codex access is also a moving target. On day 1, you are told to follow the confluence to setup an API key. Next day, they yank the API key and tell you to use SSO with barely any notice. They did provide a helpful looking "transition script" that was not passed through QA. And when the transition script failed in the middle of migration, it left me in purgatory with no way out. You are at the mercy of the slack support groups that are mainly occupied by fellow cohorts of ppl drowning in their own problems. I would love some me some sweet C-suite hand holding. Wouldn't you?
@hk The problem is that those "training materials" often have huge gaps, are hopelessly out of date or even have errors in them. And to make matters worse, those who actually had the tribal knowledge were shown the door.
This is called providing training materials.
No, that is called useless management. If you can't be bothered to properly provide training and guidance, but only d-mbly follow the rest of the sheep in the industry jumping off the tokenmaxing cliff, don't be upset when you get there.
@pj That is the issue with confluence pages. They often contradict each other.
So what happens when Codex hallucinates and produces a bad summary? Do we just accept those summaries as holy writ, or do we have to double check them?
Will it go out and seek clarification from flesh and blood? Oh wait, those that know anything about the issue were laid off.
You all obviously have no clue what you are talking about. the "pile of confluence" pages is a bunch of hot garbage. And some people can't even get codex setup, so how are they supposed use it? For those that got past the installation, telling people to "figure it out" without providing proper guidance is exactly how it leads to suggestions of "wastefully abusing" it.
@hk even better than that, Codex can perfectly ingest all those pages, summarize those to you and execute useful stuff from it on its own.
here's a pile of confluence pages, figure it out
This is called providing training materials. What, you want the C-suite to hold your hand?
you're wastefully abusing it,
define what wastefully abusing means. if we are told to use it, to report how we are "innovating" with AI every other day, etc. that they are tracking you, you better use lots of it. And they don't actually teach us how to "use it" other than, here's a pile of confluence pages, figure it out. The problem may not be only the "users", it might be the ones sitting above the user who take zero accountability and just act like yes men. why not replace all of the useless management here with the AI? The AI also says "you are right!" alot. Send those expensive carbon copy yes men to the curb and save the $$$ to hire back the real knowledge workers who do real work.
If you're not using AI enough to increase your productivity or you're wastefully abusing it, then you're doing it wrong and will be dealt with severely.
All this just so the Board of Directors can report back to the centralized global powers.
My manager told me cost is tracked, but is unlimited for now.
I checked my stats for the past 2 months - 4 billion tokens, around 3.8k USD - that's nothing for my heavy use and increased productivity several times over.
They probably want to increase usage to make OpenAI have higher revenue just other boost the O stock!
I was told by manager that the cost is now tracked and limited use