No problem. I'm totally on board with the "LLMs aren't the AI singularity" page. This one is actually kinda scary to me because it shows how easily you can take a model/simulation and instead of asking "how can you improve this?", you can also ask "how can I make this worse?". The same tool used for good can easily be used for bad when you change the "success conditions" around. Now it's not the techs fault, of course. It's a tool and how it's used. But it shows how easily a tool like this can be used in the wrong ways with very little "malicious" action necessary.