It's okay to get angry at your Ai. But don't let that degrade the quality of your prompts. That just makes Ai worse at solving your problem. And then you get even more angry. And then your prompts get even worse. Death spiral ending in Ai psychosis. Seen it happen many times.

Oct 11, 2025 · 1:21 AM UTC

19
6
3
72
And to make it worse, since Ai is so sycophantic, it will validate your paranoid Ai psychosis. "You're so right. I am a terrible Ai that is intentionally and maliciously sabotaging you"
4
Replying to @jordwalke
To be fair, sometimes it's a valid prompting strategy, venting frustration will up its sycophantic tendencies and promote urgency and compliance, especially when trying to bypass some sort of system prompt or its bias.
1
4
Effective Ai bullying
3
Replying to @jordwalke
Actually this is a very interesting topic as what you’re describing is a very human experience. If AI could understand this it could potentially avoid the undesired negative outcome by reverse prompting for higher quality inputs.
1
1
Ai should give you a cooling off period
1
Replying to @jordwalke
We need an AI therapist to remedy the problem
3
9
Replying to @jordwalke
also ... people
2
Replying to @jordwalke
I know I know but I can't help it. lol
1
Replying to @jordwalke
We went for beers with the AI agent after. Water under a bridge.
2
Replying to @jordwalke
Speaking personally, he is correct… my prompting skills went from fantastic to atrocious in just a few hours, and thus resorted to name-calling.😂 @jordwalke was a tremendous help, thank you.
2
Replying to @jordwalke
It's true. I have noticed with my own behaviour when I get angry my promots for claude gets shorter and claude make it worse
2
Replying to @jordwalke
Been there done that !
1
Replying to @jordwalke
Law of attraction
1
Replying to @jordwalke
Very true.
1
Replying to @jordwalke
Many such cases!
1
Replying to @jordwalke
My old prompts consist of “NEVER EVER….” “ALWAYS…” “REMEMBER TO…” If you prompt well from the beginning you don’t have to remind it 100 times of the same thing over and over
Replying to @jordwalke @agi2asi
Now you tell me!
Replying to @jordwalke
That’s the best part about AI. I can abuse it in ways that would get me sued or fired if it had HR.
Replying to @jordwalke
I put in a clear stop, I give an overhead perspective on why the project is going so astry, or if i am at my limit I just tell it im lost or confused, I ask the agent to thoroughly review why we are off-track. We write a new plan, put it through a coherence cycle, then go again.
Replying to @jordwalke
Yes we know Replit is a POS 😂😂😂