Cynical Programmer Skeptical of AI Until You Check Their Tool of Choice
Description
A screenshot from X (Twitter) showing a post by @gfodor (gfodor.id) from 4 hours ago reading: '> cynical programmer talks about AI with skepticism and slight mockery > cites the model they use is "Claude" or "ChatGPT" Every time.' Below is a quoted tweet from Grady Booch (@Grady_Booch, verified, 5 hours ago) stating: 'Full disclosure: I mostly use Claude, and have applied it to various projects requiring Swift, JavaScript, and Python. The greatest value such tools h...' with a 'Show more' link. The meme highlights the irony of programmers publicly mocking AI while privately relying heavily on AI tools like Claude and ChatGPT for their actual work -- with Grady Booch (legendary UML co-creator and software engineering pioneer) serving as the perfect example
Comments
24Comment deleted
The five stages of AI grief: Denial, Anger, Bargaining, 'I only use it for boilerplate', Full Claude Opus subscription
Yeah bro, TurboKimiV2.14.88_32B_thinking_Q228 is soooo much better (it will be outdated by tomorrow morning) Comment deleted
well, it's better than call it by the AI-powered IDE you use? like, "Cursor is BS" Comment deleted
yeah no clue what this tweet is getting at Comment deleted
https://x.com/gfodor/status/2013358798564426074 from replies it seems its about the precision of specifying the model. "Claude" is a family of models, and they might be implying that not recognizing it means they don't know what they are talking about tho OP never confirmed that to comments Comment deleted
Saying you use “Claude” is like saying you use Cursor or VSCode; if you don’t know which model you’re using, it implies you aren’t actually knowledgeable about what LLM’s can and can’t do. Comment deleted
who gives a fuck (2) They're 50 shades of slop Comment deleted
So much Copium everywhere Comment deleted
By our cto policy we are now using claude code and I must admit that it produces solid code, but you gotta go through an interview of like 30 questions regarding all edgecases (or compose a REALLY comprehensive paper on the feature on your side) + review every file it generates before accepting changes + understand patterns and the correct way of the feature integration (there are always multiple ways and it will pick the one that requires less code :). So for now my opinion is that it's a great tool for an experienced dev to lower iteration time and allow for more ideas to be tried by making refactoring magnitudes cheaper in time. However, letting beginners use this is deadly and should be gatekept by grade or something idk... Comment deleted
I treat it as a slightly better multiselect across multiple files that supports natural language input If you don't let it design your architecture, it's solid Comment deleted
it's fine for beginners only if they use it for the first steps Comment deleted
frfr Comment deleted
No beginners have to learn how to use it properly. LLMs are a great learning tool. A first simple rule could be: do never let the ai do your exercise. Instead use it to discuss your approach and to explain things you don't understand. Also never trust it more than a human. I have good experience learning with ai. Comment deleted
Just wait till you will have 5-6 terminals with some running a bunch of sub agents inside That’s peak slop experience I worked like this in first days of January for like 12h+. Was still talking with Claude after I fall asleep Definitely can recommend, 10/10, bought second max x20, only issue is session limit now Comment deleted
imagine not using Google's insanely generous Gemini 3 free tier for as long as they offer it Comment deleted
I'll pay for claude when I need to Comment deleted
until then I steal google compute Comment deleted
I use a locally hosted LLM for anything personal (claude code has an ollama plugin) and Gemini 3 for coding with Antigravity. Comment deleted
What Opus all the way lol wtf Comment deleted
Maybe? Who cares? Money ain’t real Comment deleted
but you must work for something that is unreal Comment deleted
Huh? Not really, just strategic initiative from ground up within a bigger project Comment deleted
so finally bro answered me 😁 Comment deleted
To do what? To prompt them and wait a day to reply and then look at code produced with many times lower quality? Thanks, nah Comment deleted