this post was submitted on 13 May 2026
248 points (97.3% liked)

Technology

84623 readers
5663 users here now

This is a most excellent place for technology news and articles.

Our Rules

  1. Follow the lemmy.world rules.
  2. Only tech related news or articles.
  3. Be excellent to each other!
  4. Mod approved content bots can post up to 10 articles per day.
  5. Threads asking for personal tech support may be deleted.
  6. Politics threads may be removed.
  7. No memes allowed as posts, OK to post as comments.
  8. Only approved bots from the list below, this includes using AI responses and summaries. To ask if your bot can be added please contact a mod.
  9. Check for duplicates before posting, duplicates may be removed
  10. Accounts 7 days and younger will have their posts automatically removed.

Approved Bots

founded 2 years ago
MODERATORS
L3s@lemmy.world
enu@lemmy.world
technopagan@lemmy.world
L4s@lemmy.world
L3s@hackingne.ws
248
Amazon employees are “tokenmaxxing” due to pressure to use AI tools (arstechnica.com)
submitted 23 hours ago* (last edited 23 hours ago) by themachinestops@lemmy.dbzer0.com to c/technology@lemmy.world
25 comments fedilink hide all child comments
 

Amazon has told employees that the AI token statistics would not be used in performance evaluations. But several staff members said they believed managers were monitoring the data.

top 25 comments
sorted by: hot top controversial new old
[–] Windex007@lemmy.world 67 points 18 hours ago (2 children)

When a measure is used as a metric, it ceases to be a valuable measure.

It's literally a software engineers job to consider and understand functions. If you tell me I'll be judged by the output of a function, I will crack it open to understand what inputs maximize outputs.

I'll burn though however many tokens you want me to. Give me hardware and a blank cheque I'll bring a data center to its knees. I'm guessing your hope was that tokens burn would translate into productivity, but that hope was misplaced and that failure is on you.

[–] SaharaMaleikuhm@feddit.org 13 points 12 hours ago

But the LLM told the manager it's a great idea, so it must be true

[–] Landless2029@lemmy.world 13 points 14 hours ago

And the only way to prevent this would have been keeping it away from managers eyes.

Only the AI team should have access to that data. Maybe a separate "efficiency" data team can crunch numbers on completed tasks vs tokens.

Never let managers or end users see a chart of token usage. That's just stupid.

This is just going to burn compute even more on LLM crap for no reason. Just burn it all down.

[–] corsicanguppy@lemmy.ca 11 points 14 hours ago (2 children)

maxxing

Faxing, taxing, relaxing. Why the exxxxxtreme x?

[–] cyberwolfie@lemmy.ml 2 points 10 hours ago

You have to maxxxx those x'es!

[–] Chulk@lemmy.ml 125 points 23 hours ago (2 children)

This is going to all collapse so horrifically once AI companies change their payment model to token usage rather than a flat monthly fee. It's already starting with Microsoft and Google. Anthropic is the last (and most consequential) nail in the coffin.

Companies are building entire workflows around AI, but they are building them under the assumption that they won't ever be charged per token.

[–] Almacca@aussie.zone 63 points 23 hours ago (2 children)

I can't wait!

[–] IratePirate@feddit.org 21 points 14 hours ago

Roses are red
AI is slop
My homies can't wait
For this bubble to pop.

[–] Chulk@lemmy.ml 36 points 22 hours ago (1 children)

Yeah I can't say I feel bad for anyone involved. But something tells me the people who created this problem won't be the ones suffering consequences.

[–] Almacca@aussie.zone 8 points 21 hours ago

Yeah, I'm sure they've got their golden parachutes ready, and will move on to the next scam.

[–] GamingChairModel@lemmy.world 20 points 22 hours ago (1 children)

Companies are building entire workflows around AI, but they are building them under the assumption that they won't ever be charged per token.

Or worse, where the AI models underpinning a workflow breaks or degrades in some way to reduce token usage and then starts behaving in unexpected ways, in a process/workflow that assumes a particular type of behavior.

[–] diabetic_porcupine@lemmy.world 1 points 5 hours ago

Sounds like you have first hand experience in this matter

[–] muntedcrocodile@hilariouschaos.com 12 points 16 hours ago

Just ask the ai for the seahorse emoji and watch it burn tokens trying to figure it out

[–] circuitfarmer@lemmy.world 21 points 19 hours ago (1 children)

There is no way any of this ends up profitable once the money hoarders stop eating their own shit.

[–] SaharaMaleikuhm@feddit.org 7 points 12 hours ago

It's gonna be so fucking funny once they actually have to make money with their AI bots.

[–] HubertManne@piefed.social 12 points 17 hours ago

I think most folks know when the company says they will not do X they will do X.

[–] sundray@lemmus.org 31 points 22 hours ago (1 children)

Misread this as "tokemaxing." And if you've ever had to work at a warehouse... yeah.

[–] cecilkorik@piefed.ca 11 points 22 hours ago (1 children)

Why not both! The more tokens you can convince the AI to spend time thinking about and spitting out, the longer you have for a toke break.

[–] paraphrand@lemmy.world 9 points 20 hours ago

slopmaxxing pro strats.

[–] gedaliyah@lemmy.world 18 points 23 hours ago

Yeah... totally normal that a new technology has to be shoehorned into the workforce. I'm sure they only need to enforce its adoption because it is so useful.

[–] bridgeenjoyer@sh.itjust.works 10 points 22 hours ago (2 children)

Can I do this at work without getting caught ? Our folks are well meaning but caught up in ai hype and have no idea how any of it works just that "its the future". They dont know anything about computers in general. I often have to explain how bad of a job it does and how useless it is for 99% of our jobs.

I'd like to waste all our tokens on it so they stop paying altman thousands a month for it.

[–] Septimaeus@infosec.pub 1 points 6 hours ago

Anyone employed as a domain expert who finds their advice consistently ignored should eventually pause and consider what else they may be inadvertently trading for their fee.

[–] PolarKraken@lemmy.dbzer0.com 3 points 15 hours ago* (last edited 14 hours ago)

You can always simply lie and say that you get great results when you send it huge amounts of text and really spend time to correct what it's getting wrong over the session.

If you can make it sound like genuine advice and enthusiasm (that you're simply wrong about), well congrats, you look like a real team player, and you cause anyone who believes you to also believe that those delicious productivity gains are just another few thousand yeeted tokens away.

Your failure to demonstrate such, should it ever come to that, can result in something like "I swear they made changes, they keep doing that silently. This used to work." Which points to a huge and real problem, while covering thine ass.

This isn't how I operate at work, to be clear, but I'm in a way different situation. Sooner your org realizes there is no mythic Shangri-La to be found, the better for all, I figure.

[–] stsquad@lemmy.ml 0 points 20 hours ago

I have API access at work because I don't want to be tied to a UI. I'm very aware of the cost because I'm trying to see where it offers good value for money.

Of course things like the deep research and notebooklm are covered by the Google workplace fees which while including more than the personal plans are also a fair bit more expensive.