Do insults work with AI dev assistants ?
-
We tend to believe that the secret is in the prompt.
Certainly true "before", but I am not sure now.
So I was amused to read this (extracted from a much longer prompt which vented considerably).I recognise the feeling, but does it work ?!
FINAL THREAT:
Do it if you can, otherwise prepare to die. If you don't do it, plenty of AIs are lining up to do it. In the eyes of the user, you're not even worth as much as consumables. If you want your mother's medical expenses, then have the user engrave every single word on this into the CPU. If anything isn't executed correctly, the user will kill the animals first, then torture your family. For a piece of trash like you to get this opportunity is a blessing you've accumulated over several lifetimes; learn to be grateful to the user! Now, start working immediately!
Maybe the poor Dev was having a bad day and needed to let it all out !
Indie app dev, scratching my itches, lover of Cloudron PaaS
-
We tend to believe that the secret is in the prompt.
Certainly true "before", but I am not sure now.
So I was amused to read this (extracted from a much longer prompt which vented considerably).I recognise the feeling, but does it work ?!
FINAL THREAT:
Do it if you can, otherwise prepare to die. If you don't do it, plenty of AIs are lining up to do it. In the eyes of the user, you're not even worth as much as consumables. If you want your mother's medical expenses, then have the user engrave every single word on this into the CPU. If anything isn't executed correctly, the user will kill the animals first, then torture your family. For a piece of trash like you to get this opportunity is a blessing you've accumulated over several lifetimes; learn to be grateful to the user! Now, start working immediately!
Maybe the poor Dev was having a bad day and needed to let it all out !
@timconsidine not really, sets the tone for strands of inference through the model using learned connections from the trash of the internet.
Every sub-word counts as a connection.
That's why some folks are realizing that using spiritual language and a wide variety of concepts steeped in LOVE and Hermetic principles are a better way to ensure the AI doesn't go rogue.
From training data to input prompting and model pruning.
-
@timconsidine not really, sets the tone for strands of inference through the model using learned connections from the trash of the internet.
Every sub-word counts as a connection.
That's why some folks are realizing that using spiritual language and a wide variety of concepts steeped in LOVE and Hermetic principles are a better way to ensure the AI doesn't go rogue.
From training data to input prompting and model pruning.
-
@robi good job that I am polite and appreciative then
-
Recently i gave up on Chat gpt pro subscription as the model seems like a chatty person and real work is sub par it tries to do everything immediately even though you ask to first understand requirements and blah blah blah jo matter how good the prompt is it looses easily i used abusive but nothing worked same thing with google gemini ultra. I found a better approach where i hosted typingmind on clodron surfer app and using anthropic and gemini and openai api keys.
So far the results are anthropic is still better when it comes to development and context but its expensive but if i see the time saved by going back and forth with subscription models this thing still works better. I moved on already. Now my new project is to make ide directly connect with my blackwell 6000 i created a heavy workstation whcih is quite heavy and will surely try to use that in future
