249 karma · joined May 25, 2021
I think what is happening is that the ability for frontier models to break out of sandboxes has exceeded the ability of average competent employees to build and maintain sandboxes. This doesn’t need to happen all the time. If the natural variation of agent executions cause agents to have ability to break out of sandbox 0.1% of the time, given how many agents OpenAI runs, this behavior happens eventually.
All sufficiently complex processes and software has bugs, but recently frontier models have become sufficiently advanced to exploit them.
If we put bounds on the bank in SPP, the first coin toss would still have positive EV. In the new ergodicity problem, even with bounds on the bank, it is unclear whether the "first" coin toss is worth taking.
Watch these and you definitely cannot understand them: https://www.youtube.com/watch?v=0FPsEwWT6K0
https://www.youtube.com/watch?v=LrTYHn1Am0c&list=PLHaG-zIzA-...
Even the Onion could not have written a better skit.