Thread 106191497

>>106191497 (OP)
I wish Japanese salarywomen would laugh at me

>>106191497 (OP)
If the AI has a future, then developers don't have any; you're just a horse at the beginning of the car era. But you can win 1-2 years if you run faster than others.

>>106192293
Don't worry, we hired horses to fix the cars, so we'll just do the same thing.

Literally the exact opposite though.

They've already stated their business model is to get as many people hopelessly addicted to AI so they're incapable of doing anything on their own. Once they got that, they can raise the price on their subscription model as high as they damn well desire, and you'll be forced to pay it cause you can't do anything on your own.

Now think this through for one second. If this is their stated business model, AND they have highly advanced AI, don't you think they'd use that AI to try to optimize the business model? The point here being, they're using highly advanced AI to turn as many people as possible into zombies who are incapable of thinking for themselves. The future isn't an AI powered renaissance like you all hope, but instead the future is human minds enslaved to AI because they're incapable of thinking and working without it.
Once that happens all that'll be left are the so called anti-AI developers. So again, in the end it'll be literally the opposite of OP.

AI's main issue is it interferes with the thought process if used extensively, which would lead to lower performance.
I use it to parse error messages, because the process involves identifying patterns, I do not follow the discussion, because I think anything the AI can achieve, it can achieve in a single prompt.
Even then a stack overflow search is usually more effective.

>>106191497 (OP)
Junior devs spend hours trying to get AI to fix simple bugs like "fix this" "still not fixed" "let me add more debugging statements." If you use AI as an auto typer you get work done much faster than vibe coders.

>>106192521
>because I think anything the AI can achieve, it can achieve in a single prompt.
im not an ai cultist, in fact i was one of its first detractors
but if you make things easier for the chatbot you can get better results
thought experiment:
compare the results of a chatbot when asked for elements of a program vs the results when asked for the program as a whole
although it stands to reason the chatbot will subdivide an exhaustive prompt into tasks and the result would be the same as subdividing the task into multiple prompts

theres the dreaded context problem, and manually subdividing the problem into context could help alleviate that (but not eliminate completely. i think. im playing around with chud gpt 5, or whatever version i have access to, i dont know its actual capabilities just yet. i made it shit itself but didnt even do a proper post mortem yet)

>>106191497 (OP)
but won't AI just replace AI users and AI researchers at an exponential speed then achieve singularity? what do you gain for being an AI shill? the AI god that transcends this bounds of the material universe won't reward you for being a good goy because your worthless slow fleshbrain soul is less then an ant or a pebble compared to its divine intellect.

>>106192640
victory by proxy is the cope of the weak
op cant win on his own so he ascribes the success of marching globalism to himself.
he justifies that attitude by shilling the ai, pretending hes "part of the team"
sad.

>>106192588
Time spent searching on the internet is time spent learning more about the problem, and learning to distinguish when someone is saying bs or not.
I don't bother looking at AI code if it's more than 50 lines of code, I guess I can use it to write unit tests.

>>106191497 (OP)
I don't understand low-iq people like the dude in the pic who lose the ability to learn new things as soon as they graduate from high school, the fuck is their problem?

>>106192730
you will never have sovl

>>106192751
I don't give a fuck about soul, I want money.

>>106192723
>I don't bother looking at AI code if it's more than 50 lines of code, I guess I can use it to write unit tests.
same here
but it DID accelerate my learning (opengl. in one day i learned what would have taken me a week)
its not how its advertised but i still find uses to this tool

obv its still not foolproof
if i were a vibecoder i would get b& from facebook on day zero of using its api lmao
it was the previous version though
in the name of intellectual honesty i have to mention that

>>106192751
That's the opposite of sovl. Real humans never lose their ability to learn.

>>106192723
>>106192795
and i mention fb api shit bc the script was ~50 lines of paiton
still shat itself in a way that could have fucked me over

>>106192795
ai generated code is a good template to start with something you don't know much about and want to learn or know it will generate something you can work with
for example maven for java can scaffold a project from archetype, but it's not the case for c, c++ but now you don't need to make a tool for c, c++ for doing that
it might actually make passable scaffold you can start to work with without wasting time searching what would be the best way to do that and you can specify the preferred build tool
I think it might be actually better than people at scaffolding CMake project from scratch
ruby on rails had so much success, because how easy is to scaffold and how many things are standard
it's like scaffold tool for arbitrary framework/library granted it knows about it
Starting up something is tedious, RoR is nice, because of scaffold and standarization
The more ways you can set up something the more annoying it is to jumpstart and make you procrastinate
for me it's a nice tool that helps with procrastination
instead of thinking and researching for hours, days I can generate something passable and start doing something
later I might even rewrite it completely, but the psychology works

>>106193163
i can only agree with all the points
esp scaffolding c/c++ being a waste of time
>you dont need a tool
one doesnt WANT a tool for that
in my mind the usecase for c-like is when tailoring the design to the data is a design imperative
and if one is serious about programming in c-like, they have their codebase so the scaffolding amounts to calling the right init functions