I know right, haha. But to be fair, there’s something to this. We train these models on a bunch of data and see what emerges, kind of like cooking or some primitive form of chemistry. So maybe it’s not so silly to interview the model a few times and average the responses haha.