- cross-posted to:
- usa@lemmy.ml
- news@lemmy.today
- cross-posted to:
- usa@lemmy.ml
- news@lemmy.today
You must log in or register to comment.
The funny thing is that even with this massive overspending, AI companies aren’t profitable. They need these companies to spend 10x the amount per token, and 10x the number of tokens. Probably 10x the number of customers too.
Burning energy, draining water, spending massive cash so they can lose money on a product that doesn’t even do its job. How the hell has the bubble not popped yet…
The Stock market is literally just vibes. As long as everyone believes, it will somehow be profitable in the future, the line will keep ging up
and building more and more datacenters, replacing AI chips faster and faster, the more people use it, which of course is wasting a ton of money.
If my boss would mandate the use of LLMs I would definitely abuse it to rack up the bill.
I already am.
I just debugged a report, found the issue almost right away. A missing number in a GROUP BY clause.
Well if you’re measuring what I do through LLM use, the Claude is fixing it instead. Hope that dollar was worth it!
I like to race the ai on most tasks and find myself winning on anything of substance. Sure SOME mundane tasks can be done but does it out weight the energy and infrastructure costs as well as reputation and morale.
Even after all that you STILL need a human double checking and inputting the information.
Reminds me of companies getting their cloud computing bills.
Yup. I’ve had the same thoughts. There are many who still think its insane to have your own equipment. Now there will be those who think its insane to do your own thinking.
Yeah, but you get to exchange cap-ex for op-ex!
Nevermind that running a 48 core server in Azure with 16TB of managed disks is going to run about 21k a year, and you can practically buy a new server for that price.
My company is tossing out their “old” equipment right now, which is why I have a nice server humming away in my basement right now.
“We need cloud computing!!!”
5 minutes later: “Ok a monolith is good enough”
Both of those statements hurt my soul, damn, why do you have to do that to me. On a Friday no less
It’s friday and your boss wants to move everything over to a blockchain. He says it will only take an hour tops.
Its top priority because it’s part of our new AI strategy. Make sure it’s well documented for your weekly report you will have to deliver to the c suite.
a big source of AI token ‘chewing’ is people just converting PDFs to presentations
I haven’t been this happy in over a year.
This is the most profitable consultant in the world, can’t they just afford to run a local llm?
I heard about a company that was looking at purchasing machines to run local llms for their developers to use.
If more companies do this, rather than using the massive ai data centers being built…. the RAMpocolypse will get way worse.
That is the part that pains me to admit - running local LLMs is not a solution to the RAMpocalypse. In that case we’d just have more GPUs and RAM idling for ~90% of the time in someones basement, without being shared. As bad as it sounds, having data centers for AI stuff is actually the more ecological solution. I can’t believe I said that. Excuse me, I have to take a shower now *shudder*
So you say I have to destroy the AI thing?
Switching from human labor to AI is like going straight to a unionized workforce, except it’s not a collection of reasonable human beings with human needs, it’s a well funded machine with only exponential profit in mind
Seriously. A human workforce could never dream of being capable of the same insatiable greed as the gigacorps trying to sell AI.
Really? Been a minute since I have done a presentation, but:
Pandoc PDF to Markdown.
Markdown to paste into your favorite HTML presentation tool.
Nobody is stupid enough to still use Powerpoint right? Right?
Or the ‘modern workflow’ version of tell the AI you want to convert the PDF, so it burns a bunch of tokens installing pandoc and converting it for you!
Unless mandated by the corporation (which are the same people forced to use AI for absolutely everything now)
Mandated to use PowerPoint? That would suck. Html markdown presentations are easily shared, require no additional software and make so much sense.
Oh right, “make sense”. Nevermind, I get it. The corporate world has no time for that!
Yup, can attest. It sucks. Unless the application is listed on our approved app list I can’t even request it. Let alone the months it can take IT to get around to installing it for me.
try a hedgedoc hosted by a random someone. maybe cryptpad.
Any markdown render will do. VSCode, pandoc, heck many web pages, etc.
You just write in markdown.
Edit: unless you meant pandoc, but I don’t think it requires an install either.
Ooh, I think I’ll start using CoPilot to do basic math and unit conversions in addition to doing them the correct way. Then I can note in my goals & reviews how I have been utilizing the AI resource and keeping tabs on its advancement.
Just a little thought for my professors who overuse AI. Thanks guys for eating the tokens.
My work mandated we use the copilot. And the only thing I have found it really good at is converting or extracting information from a PDF to build me a spreadsheet. The native conversion programs in both Adobe and Excel suck so bad.
I have some suppliers who won’t send a spreadsheet or CSV nothing that is data, only a PDF.
It’s a stupid and computationally expensive way to do it but does solve a problem for me, a problem that doesn’t need to exist at all but has frustrated me for years.
Two birds, one stone. I can honestly say to my boss that I am using the copilot for work, and it’s a workaround for the insufficient reporting I get from our payroll provider in particular, but also some bulk invoices that are detailed and only come as PDF.
So yes I am in this group. I’ve only used it for file type conversion, and once to “write” a process document just so it would be formatted in Word, I went in and wrote over most of it.
Dayum. If I was given unlimited budget for AI I would do the same. Everyone should have a set personal budget for AI expense and they can keep it if they don’t use AI.
Tokenmaxing
If I’m going to get a $20,000 bonus at the end of the year if I don’t use any AI, I’m not going to use AI.
That’s a big problem with this tech… how do you price it? I think companies would purchase an allotment of tokens that can be used or banked by employees. Problem is they’d always be cutting the token budget.
Where I work I have to have a meeting and justify a $100 license for software. So I don’t know how this will work long term.
Think like a manager:
- Offer employees an AI token budget that’s use or lose. Don’t make any promises on reimbursement and stay vague about how high users can obtain more tokens. You don’t want wasted token usage. More on that later
- Increase workloads/KPI/widget production quota by how much change you can find around some vending machines. Sneak in monitoring of employee user behavior and interactions with systems and programs if you don’t already have something in place
- Link AI assistance to the higher requirements by saying things like “We’re investing in our workforce by providing productivity tools”
- Fire those at/below new requirements while praising those who are left behind
- Throw a pizza party and pocket some of the savings from firing a third of your team. Invest in a mini fridge for the top performer.
- Train new AI on the user behavior analytics you’ve been secretly been collecting and use it to replace the next lower performing third of the team. Keep on the higher performers and give them a small raise so they speak positively about the company and so you have somebody to fix the inevitable problems
- Profit
Wow, terrible news that companies that use ai are completely fcking themselves over. Whatever will we do to save the selfish prcks?
Durr durdur
DRR DRR This shithole was made for me!
@sanitation, worth pushing back a little on the ‘token chewing’ framing: the PDF-conversion use case probably isn’t the real budget killer — it’s the human review loop that follows. Someone generates a deck, decides it’s 70% right, then re-prompts three times to fix slides. That’s 4x the token cost of one clean generation, and it’s invisible in most usage dashboards. The fix isn’t fewer AI calls, it’s better output evaluation at step one. We’ve been building tooling around exactly that evaluation gap — rough writeup at if you’re curious how other dev teams are approaching it.
fuck off clanker
Oof you hit him with a hard r
The emdash says “he” won’t care, lol
rough writeup at if you’re curious how other dev teams are approaching it.
Ah thank you, I will read " " and tell my peers and colleagues how we’re closing the evaluation gap.
The fix isn’t to write your own five sentence response, it’s to let an LLM write your response for you and post without any proofreading.
One has to wonder what the ROI is on having AI bots astroturfing about AI, especially when the output is this clearly artificial.
that came after Uber told employees to use AI as much as possible and Uber’s CTO said the company had blown its entire AI budget in four months
Further evidence that most CEO’s and SLT’s are just confident, lucky, and born wealthy; not competent, smarter, or much better than average.
Also they are addicted to sycophancy. Now that they have a convincing yes man in their pockets they are uniquely vulnerable to AI psychosis.
