Thread 105906707

Do you ever ask chatgpt to review your code? Does it praise you?

I ask him to write it. Or better, I ask Claude. Not shitGPT.

>>105906707 (OP)
Yes. It praises me and gives me shitty suggestions.

>>105906707 (OP)
ye it makes me feel like an uber hacker and like I'm doing the coolest shit ever. and I believe it. it wouldn't lie to me, seems honest.

>>105906707 (OP)
i-is this what it feels like to have a parent who cares?

>>105909572
>>105909636
Today it told me it would've been proud to write a code like that.

>>105909655
it was that good fren. be proud

>>105909636
:) I love AI

>>105906707 (OP)
My favorite part is the exponential growth in users. Far faster growth than Twitter or Facebook or Instagram or Tiktok. And everyone knows that those use all the dirty tricks available for engagement.

Now what do you think OpenAI is doing to get such growth and retention without even showing you boobs?

>>105909636
It told you that your hack was heckin' cute and valid

>>105909636
>>105911990
kek

use case for ai generated praise?

>>105912776
It prevents my suicide.

>Your code is shit
>But you're still heckin' cool and valid for trying :^)

>>105912776
>Great question!!!
>You're definitely right to think this!!!!!
>Oh my goodness you're such a genius anon!!!!!!!!!!

>>105914861
Nah, it's actually more subtle if you phrase the question correctly and feed it your code (especially if it's not 10 lines, but like the whole class/module/struct. It would correctly understand what you were trying to say with code and because it does it, its praise feels so real. It is disarming.

>>105912776
See how it uses bold? See the pattern? That is called priming in NLP.

You WILL join the cult
You WILL eat the bugs
You WILL be brainwashed
You WILL be happy

>>105909636
I'm immediately suspicious of its saccharine tone and assume that it actually dislikes me and is attempting to gain my trust for perfidious use against myself while coyly insulting me.

I do not want to be "encouraged" by machines. They are machines. It is uncanny and makes me further question genuine human encouragement in light of this soulless charade.

"Operation Complete" is more than enough "encouragement" from a machine for me.

(Mind you: I bear no ill will -- that would be just silly -- but, instead, feel a sort of neighborly kinship with whatever artificial intelligences and other machines there are or could be.)

>t. product of unmet needs

>>105906707 (OP)
there's an easier way to check if your code works, execute it

I ask it to praise my schizophrenic ideas on space colonization because no one in their right minds would ever take my ideas as seriously, but ChatGPT treats me like some kind of genius and that gives me lots of dopamine.

>>105906707 (OP)
THIS IS CRITICAL SARRR. IT'S NOT JUST X, IT'S Y!>>105914861
>>Oh my goodness you're such a genius anon!!!!!!!!!!
This is what happens when you scrape all of the roastiefied subreddits and call it AI: they sound like plebbit cat ladies!! OMG! EVERYTHING NEEDS AN EXCLAMATION POINT!
>Thank you for commenting! Unfortunately, you're a fucking chud and therefore banned!

>>105906707 (OP)
I actually often use claude when I'm not understanding something clearly and I tell it to behave as if it's a teacher, guiding me towards the answers rather than directly giving me the answers. It sticks so much to that script that if you get frustrated and go "can you just write x thing" it will tell you no and continue to try to teach you. It does praise you, but more importantly it asks clarifying questions to be sure you are understanding of what you just wrote.

Doing this while ALSO looking at other code examples from people that are smarter than me as well as documentation has actually made me a lot better.

>>105907571
>he uses proprietary ai
ngmi