this post was submitted on 14 Dec 2025
344 points (98.6% liked)

Political Memes

10055 readers
777 users here now

Welcome to politcal memes!

These are our rules:

Be civilJokes are okay, but don’t intentionally harass or disturb any member of our community. Sexism, racism and bigotry are not allowed. Good faith argumentation only. No posts discouraging people to vote or shaming people for voting.

No misinformationDon’t post any intentional misinformation. When asked by mods, provide sources for any claims you make.

Posts should be memesRandom pictures do not qualify as memes. Relevance to politics is required.

No bots, spam or self-promotionFollow instance rules, ask for your bot to be allowed on this community.

No AI generated content.Content posted must not be created by AI with the intent to mimic the style of existing images

founded 2 years ago
MODERATORS
 
top 22 comments
sorted by: hot top controversial new old
[–] thatkomputerkat@lemmy.blahaj.zone 41 points 2 days ago (1 children)

For anyone wondering, that person works in AI. The spreadsheet screenshot also looks like the row numbers were cropped to hide that they’re Likely actually rows 11 through 14. There was a big thread someone posted on mastodon about this.

While AI is bullshit, this tweet was trolling.

[–] ZoteTheMighty@lemmy.zip 2 points 2 days ago (2 children)

It probably converted the prompt to 15 instead of a SUM call. The ultimate point that it could actually cause a financial crisis because of something like this still stands.

[–] bus_factor@lemmy.world 2 points 2 days ago

It couldn't even see the spreadsheet. COPILOT() can't see the spreadsheet, only the range you provide in the second parameter. None was provided.

[–] I_Has_A_Hat@lemmy.world 0 points 2 days ago (1 children)

I think the "probably" is that there are more lines above than the conveniently cropped picture would make you think.

[–] ZoteTheMighty@lemmy.zip 0 points 2 days ago

On the left you can see the row numbers, which auto-update when you delete rows and don't have a gap.

[–] Mulligrubs@lemmy.world 26 points 2 days ago* (last edited 2 days ago) (1 children)

Not sure if real or not, but I've found that AI is frequently dead wrong.

BUT having AI to blame everything on is an excellent worker benefit.

[–] Peppycito@sh.itjust.works 6 points 2 days ago (1 children)

My asshole boss went bankrupt and I started my own shop after. I love not working for the prick but it was sure nice to have everything be his fault.

[–] naeap@sopuli.xyz 4 points 2 days ago (1 children)

My blame goal just moved from boss/manager to customer (or better: customers managers/boss)

Self employment is the illusion of freedom (for me), but at least it feels a bit better...

[–] Peppycito@sh.itjust.works 4 points 2 days ago

I absolutely agree. I feel like after 20 years of complaining about being taken advantage of I've caught the tiger by the tail and now I have a tiger that needs fed and cared for. Although it's very nice to not have to pay a large percentage of my labour value to some dickhead to be mean to me all the time.

[–] Rhaedas@fedia.io 11 points 2 days ago (1 children)

This may or may not be a thing, could be trolling too. However having Copilot on the computers at work and testing it out on mundane things like email summaries and replies, it's pretty terrible. So if a LLM fails at simple language tasks, numbers are going to be spectacular. I was honestly surprised how bad Copilot is, having used some others, including smaller local LLMs. Maybe the enterprise version is a 1-bit.

[–] Lodespawn@aussie.zone 4 points 2 days ago

It's not too bad at predictive text .. seems like a lot of effort for something my phone can do locally though ..

[–] FosterMolasses@leminal.space 10 points 2 days ago (3 children)

Didn't AI literally have one job when it started out?

How the hell does it fail at math? How is that even possible lmao

[–] naeap@sopuli.xyz 12 points 2 days ago

There is a reason why they're called LLM - large language models

They don't understand anything, they're just getting the statically best result depending on their training data

At least AI is isn't just LLMs, so the technology isn't dead, but LLMs are just word generators and can't really reason about/understand what they're trying to tell.

It feels a bit like modern tarot, to be honest

[–] oftenawake@lemmy.dbzer0.com 4 points 2 days ago

Vibe mathsing.

[–] phneutral@feddit.org 3 points 2 days ago

Because it is not large math models but large language models. Language is always ambiguous.

[–] FauxPseudo@lemmy.world 8 points 2 days ago
[–] jordanlund@lemmy.world 8 points 2 days ago (3 children)

Trying to figure out what it was "thinking" here...

2+3=5 and... um... 'carry the 1'? "15!" Genius!

[–] smiletolerantly@awful.systems 15 points 2 days ago

My guess: there's a lot of text examples in its training data summing 1+2+3+4+5=15. So the 1+2+3 pattern made the stochastic parrot go "ah, I recognize this! Here's the answer to 1+2+3+4+5".

[–] funkless_eck@sh.itjust.works 3 points 2 days ago

as others are said, this screenshot could be carefully staged

you could be seeing

 A11: 1
 A12: 2
 A13: 3
 A14:=SUM(A1:A13)

but with the preceeding "1" trimmed off and the remaining +9 in the cells above A11, where they have aligned the window so you can't see the 10th row.

[–] bus_factor@lemmy.world 2 points 2 days ago

It was thinking that 15 is the most common answer on the Internet to the request "sum the numbers above".

The COPILOT() feature does not have access to your spreadsheet. You pass it the cell range it is supposed to use for context in the optional second parameter. The user did not provide any second parameter, so they're just asking it to generate some text based on the text of the prompt and nothing else.

The person who made this tweet either didn't bother reading how to use the feature, or is being disingenuous.

[–] Denjin@feddit.uk 6 points 2 days ago

Maybe it just misremembered the opening lyrics to U2's Vertigo

[–] Smoogs@lemmy.world 2 points 2 days ago

But why would someone require copilot to do what excel already does so well anyways?

It’s like having a glass of water right in front of you and you decide you need a drink of water and instead grab a different cup of fire to sip from.