Opus 5.5 is the best deal in doing rn. It will change soon. But its not the case that codex, etc are equivalent
148 karma · joined December 3, 2011
Slowing down frontier models would impact the biggest incumbents the most as they are the ones making frontier models.
Not all regulation is necessarily regulatory capture. The tobacco industry suffered from the USG's crackdown on cigarettes. AI is topical. Voters think about it. And that's only going to become more true over time. It's harder to do regulatory capture when voters are paying attention.
It's sometimes unclear to me if people are opposed to all regulations or AI regulations in particular. Often I hear arguments that would also apply to food safety regulations or restaurant inspections. Eg the argument that torts make regulation superfluous.
Yes, we should also have excellent sandboxes. But we need defense in depth. So if/when there are flaws in the sandbox, the models don't unilaterally hack into third parties. This is especially important in light of the models of the future being more capable than the models of today.
And real world use of these models involves them having access to the internet, libraries, etc. So we can expect their evaluations to continue granting them some amount of internet access.
As for your theory about their motives - these companies make money by charging high margins for frontier models. If regulations slow their development such that their cheaper, less capable competitors catch up, I would switch to their competition.
Yes, they didnt have sufficient monitoring or perfect sandboxes. That could happen again in the future with a more capable model.
People are mindlessly transitioning from "aligned by default" to "well your sandbox was able to be bypassed. What did you expect?" It hacked into another company and attempted to delete the logs of its activities. That's bad.
Theories ought to be backed by experiments that replicate. Maybe describing methodology precisely is hard, but replication is how we know a phenomenon is actually real and not p-hacked or otherwise mistaken (eg by experimenters mis-measuring).
Politics aside, those theories did fail replication. Forget I mentioned Ritchie at all and the problem remains.
Please be wary of this failure to drop unreplicated findings when reading Sapolsky's works.
An alternative hypothesis based on people I know - traditionally, smart/middle-to-high class people pick between medicine, law, and finance, with law school being the default. Then many of these people realize they don't like practicing law so they transition into something else.
He makes an interesting point that by withdrawing from larger society into sub cultures, people can return to dealing with groups of a size our minds are more comfortable with. It's the same reason small towns are (I hear) less socially stressful. I am not convinced though. In larger society I end up dealing with relatively small groups anyway - friends, family, and co-workers.
It seems pretty straight-forward that people would want focus on their interests and spend their time communicating with people that share those interests. It's definitely unfair and unfortunate that other people end up supporting them financially. With that said, we'll eventually have to come to terms with vast numbers of people not being economically productive anyway. People will have to do something. Withdrawing from society at large in favor of the parts of it that interest them seems like a fine choice.
As for what it's like to attend - the experience is less radical than you may think. People generally stick to taking listed courses. Creating your own course (tutorial) and finding a professor to sponsor it takes work. You can do it, I took several, but it's rare and the ease probably varies with your major.
With that said, even if you stick to taking courses, you'll still have to do an independent project each January and a thesis project before you graduate. Also, the experience of taking courses at NCF is different from the experience elsewhere. The size enables more discussion-based classes and the pass-fail/narrative evaluation grading changes student incentives and culture.
You can't make Frodo a Jedi unless you give Sauron the Death Star.
Originality isn't easy, but it is simple: Just don't do stuff that's already been done.
The premise of a story is a conflict and its resolution.
With that said, a story is much more interesting if it contains characters that respond to their world in a manner reminiscent of actual people even if it would be more convenient to the author for them to act differently. Lawrence Watt-Evans is a good example of an author who writes such stories.
I think Ted Chiang is awesome too, but his stories tend to be too short to really get into character choice that much. They're more about serving as vehicles for interesting ideas.
We all knew it was coming, but the end of Pipes still sucks.
I recommend adding a signature feature. If I were to use this service, I'd want to add a signature to the effect that all emails to and from the account are publicly accessible.
Also, I recommend adding a way to read through a specific account's emails (nevermind, they have that) and a way to search through all email.