Tokenmaxxing needs to stop

Companies are starting to mandate the use of AI with their employees by using a token usage leaderboard. So the employees who use the most tokens are going to be at the top of the leaderboard, and they're going to get rewarded for burning all of the company's money. And this is obviously such a horrible way to measure productive AI usage. It's actually such a horrible way to measure AI usage that we literally have employees who are running throwaway prompts just to inflate their AI usage and climb the leaderboard. Meta, the company behind Facebook, has an internal leaderboard that tracks the token usage of every single employee, right? Their workers are just producing so much throwaway, wasteful slop trying to climb the leaderboard meta that Meta, just in the last 30 days, if I'm not mistaken, has used more than 900 million dollars' worth of LLM tokens. That's insane. Now, I think they've removed that leaderboard since then, but it just goes to show you how incredibly dumb some of these companies are. Now, Microsoft also has an internal dashboard showing the AI usage of every single employee, and I'm pretty sure they haven't actually removed it yet. It's just such a stupid way to track whether your employees are productively using AI or not. It's probably even worse than tracking the lines of code of every single employee. Like, it's just so wasteful. There's employees in these companies who are literally producing work, producing code that they don't intend to use at all, that they just throw it away, right? But it inflates their token usage numbers, and so they keep doing it so they can climb that leaderboard. Now, Salesforce is also mandating AI usage, and their way of doing it might just be the worst of them all because they require their employees to spend 100 bucks on Claude Code and 70 bucks a month on Cursor every single employee, every single month, all coming from the company's budget. And if you don't spend those minimum amounts every single month, you're gonna get flagged by their system, right? So obviously, these employees are worried about their job security. They don't want to get flagged by this system for using too few AI tokens. So what are they going to do, right? They're going to send useless prompts. They're going to generate code that they're not going to use and just do all sorts of other wasteful token-maxing practices just to inflate their token numbers, right? And the only real winners here are the companies actually selling the AI tokens. Everyone else is just getting screwed over. The only ones who are winning over here are Anthropic, OpenAI, and Google. All these other companies are just burning massive amounts of money on inference costs to produce slop that isn't even getting shipped to users because it is only created to inflate the token usage numbers so the employees don't get flagged by this flawed system. Like, the whole thing is just a big fucking mess. As much as I love AI and the problems it solves, we are very much living in a big bubble right now, and I would be very happy once it pops. That's all for this video. I just found these companies doing this dumb thing, and I just had to talk about it. I'm sure you guys have some thoughts of yourself on this topic as well, and I'd love to see them in the comments below. So drop it below. Other than that, thank you for watching, and I'll see you in the next video. In the meantime, there's going to be some other videos up here on the screen that you can go watch as well.