Google Gemini Model Picker Labels 3 Pro as 'lmao' Instead of 'New'

Description

A screenshot of Google's Gemini AI model selection dropdown on a dark-themed mobile interface. The header reads 'Choose your model'. Two options are shown: '3 Pro' described as 'Cutting-edge intelligence' with a blue pill-shaped badge reading 'lmao' (instead of the usual 'New' badge), and '2.5 Pro' described as 'Reasoning, math & code' with a standard blue 'New' badge. Below the dropdown, the currently selected model shows '2.5 Flash' with a dropdown arrow and a microphone icon. The humor comes from someone editing the 'New' badge on the 3 Pro model to say 'lmao' instead, mocking either the model's capabilities or Google's marketing of it as 'cutting-edge intelligence.'

Common Tags

Google Llm AIHype AITools TechSatire AIHypeVsReality TechIndustryIrony MobileDevelopment DarkMode UIDesign

Comments

Anonymous ★ Top Pick When even the UI badge can't keep a straight face about your model being 'cutting-edge intelligence,' you know the benchmark scores were written by the marketing team

Anonymous ★ Top Pick

When even the UI badge can't keep a straight face about your model being 'cutting-edge intelligence,' you know the benchmark scores were written by the marketing team
Anonymous

My ideal model selector: 'Does what the docs say' with a 'Finally' button, and 'Hallucinates a new sorting algorithm' with an 'I'm feeling lucky' button
@NaNmber 10mo

for free!
@ZmEYkA_3310 10mo

I know :(
1. @NaNmber 10mo
  
  what a banger memes you post holy lmao
  
  @linuxhasan 10mo
  
  I agree
  
  @ZmEYkA_3310 10mo
  
  ty lmao
@advanced_name_1 10mo

Lmao
dev_meme 10mo

Oh ffs
1. @RiedleroD 10mo
  
  I don't understand. what's this about?
  
  dev_meme 10mo
  
  It was about Gemini 3
  
  @RiedleroD 10mo
  
  ok but what's the context here
  
  dev_meme 10mo
  
  Well, there was a lot of hype about how good it is expected to be
  
  @RiedleroD 10mo
  
  oh ok
  
  dev_meme 10mo
  
  https://t.me/devs_chat/176513 This screenshot is about Demis, not about Elon, just to make it clear Just coz he isn't one to make hype out of nothing
  
  @RiedleroD 10mo
  
  fwiw this does seem to be a true OG in the AI scene, going back to alphago and such. people like him I do have some respect for
  
  @RiedleroD 10mo
  
  still, I don't believe it's going to be anything revolutionary. just yet another marginal improvement
  
  @DerKnerd 10mo
  
  yeah probably, and one main purpose I have for LLM assistance is just converting between languages, creating CTEs from JSON data I have to map arrays of objects to sql upserts and for that most models work more than well already
  
  dev_meme 10mo
  
  12 marginal improvements over the year year across the market is pretty good tho
  
  @RiedleroD 10mo
  
  we've gone from "obviously wrong" to "wrong in ways that are not so obvious" to "technically sound in isolation, but WILL cause problems in context of the rest of the codebase"
  
  @DerKnerd 10mo
  
  I don't even think the main issue is the coding part, but that the LLMs get better with time in being convincing even if they talk bullcrap
  
  @RiedleroD 10mo
  
  yea I mean part of getting better at convincing you it's good is actually making it good, but still it's not like an actual programmer thinking about how the code is going to be used down the road and such and such. and the amount of epiphanies I've had while working on code, where I completely restructured the thing I was implementing because I figured out the way I wanted to do it was bullshit, … yeah
  
  dev_meme 10mo
  
  >> completely restructured the thing I was implementing because I figured out the way I wanted to do it was bullshit, But when people find out the way AI goes is bullshit they go crazy and curse the LLM, instead of doing what they would do if they worked without it! Why so?
  
  @RiedleroD 10mo
  
  …because the AI won't tell them that, and you have to have a deep understanding of the codebase to figure out that you're attempting bullshit
  
  @DerKnerd 10mo
  
  the reason for that, from my point, if the ai creates bullcrap i don't know the reasoning or thinking behind it. When I do it myself I do know how I got there
  
  dev_meme 10mo
  
  Yeah, likelly, but this is also true with, for example, coworkers That meme "managers were always vibecoding" not an actual meme, it was more like actual fact
  
  @DerKnerd 10mo
  
  you do know, I can talk to coworkers to get their reasoning behind it right?
  
  dev_meme 10mo
  
  Why wouldn't you talk to LLM then? Sonnet/GPT-5 are pretty well at explaining their reasoning, and you can "catch" their hand to correct direction of work
  
  @DerKnerd 10mo
  
  I actually made horrific experience with that approach, it just starts hallucinating crap at some point
  
  @pdsnrc 10mo
  
  for some reason I just imagine someone very pissed about hiring a schizo programmer
  
  @DerKnerd 10mo
  
  oh boy, that might be interesting, coding as a person with splitpersonality disorder, one personality only codes in perl while the other won't touch anything but assembler
  
  @RiedleroD 10mo
  
  tell me about it 💀 I fucking hate being plural
  
  @DerKnerd 10mo
  
  well I had voices but they were mostly a code review you really don't want to hear
  
  @RiedleroD 10mo
  
  bruh
  
  @RiedleroD 10mo
  
  "sorry for the downtime, friedrich hallucinated passing CI and pushed to prod"
  
  dev_meme 10mo
  
  This is "person-in-the-loop" responsibility to catch those, we used to call them programmers before 🌚 The same thing was always happening, thus "minimum approves" requirement for PRs
  
  @RiedleroD 10mo
  
  …you only know the codebase if you touch a good portion of the code yourself
  
  dev_meme 10mo
  
  Or if documentation is well structured, and domains of responsibility are reasonably separated
  
  @RiedleroD 10mo
  
  I will say, AI is probably ok for generating docs. and no, reading docs is not a substitute for knowing the code
  
  dev_meme 10mo
  
  And it being SOTA on lmarena preliminary confirms it but personally I will need a few days to play with it
@RiedleroD 10mo

very funny. I love masturbation
@RiedleroD 10mo

not officially diagnosed, but yes

Use J and K for navigation

Google Gemini Model Picker Labels 3 Pro as 'lmao' Instead of 'New'

Description

Categories

Common Tags

Comments

More like this

US CEOs on the Breaking Bad Deathbed After GLM Flash

Sundar Tells Gemini: Come On, Commit Crimes

Hope Gemini 4 Finishes One Day Too