This is a special post for quick takes by leogao. Only they can create top-level comments. Comments here also appear on the Quick Takes page and All Posts page.
https://en.wikipedia.org/wiki/Greenspun%27s_tenth_rule
But for real, there is a reason why even type safe languages are not truly universally type safe. Most sufficiently complex software systems will have some internal modularity, scripting etc which then does in fact enable you to spin up a completely different program inside of the program. I agree it is a sliding scale between print("hello world") and this kind of thing tho.
4Cole Wyeth
We could maybe find a “blatant“ secret behavior, but not necessarily a hidden one. My impression is that this is how many cyberattacks work, the AIs mention Easter eggs, back doors, Trojan horses, logic and time bombs, …
this feels like it’s letting perfect be the enemy of good / drastically underestimating the gap between how much we understand programs and how much we understand neural networks. like sure you need proofs to really be sure it has no bugs/vulns, but also like man you can kind of just look at programs and mostly understand what they’re doing! like it would be extremely hard to hide a copy of the game Doom inside the non-blob part of the Linux kernel source, whereas the neural network equivalent is way easier.