Cheeky question, cheeky answer. I love it. I think what I see among AI practitioners is very little knowledge of what happens next. Even the folks sitting in the frontier labs are surprised by things that happen every single week. And given that, I think the thing we should think about stopping is planning large multi-year bets around things — and instead do lots of experiments, and just be okay with experimenting. Just try stuff. Because some things will work for us and some things will be a terrible idea and then we’ll stop doing them. But because we don’t know how this unfolds — and even the people who are making the models don’t know how it unfolds — it is dangerous for us to guess, oh, in ten years the AI will be able to do this, and that’s what we should shoot for. We don’t know. We don’t know next year. We don’t know the next six months. And so therefore we should try stuff. We should be okay. And I would encourage all of you to also try stuff. At the AI pre-conference there was an AI source verifier — it wasn’t perfect; it was great in some ways, and other people were like, wait, it’s not that good in this way. But the idea of just experimenting—