I broke ChatGPT today! Got to the bottom of the well.

Complexity explorer. Residing in the liminal.

Managed to rebuild the output from the prompts, but goddam it LLMs are stochastic! It's lost some of the original qualities.

Stock picker and skydiver. I trade and dive.

@mativusgf I'm running GPT-4o, it's good but still memory constrained. A very long chat exploring the interplay of philosophy and psychology pushed it over the edge!

Interesting! How did you push its limits? Perhaps it's time to test GPT-4 and see how it holds up.