This would mean that a hypothetical AI "uniformly" gaining capability on all axes would beat us at math long before it beats us at deception.
I'm pretty skeptical of this as an assumption.
If you want an AI to output a useful design for an aligned AI, that design has to be secure, because an aligned-but-insecure AI is not stably aligned, it could be hacked. Ergo, your oracle AI must be using a security mindset at superhuman levels of intelligence. Otherwise the textbook you'll get out will be beautiful, logical, coherent, and insecure. I don't see how you could make an AI which has that level of security mindset and isn't superhumanly capable of deception.
I've been attempting to use IFS for years without having read much more than brief summaries of it. This post put me on a much firmer footing with it and I was able to much more clearly categorize a bunch of things that have been happening over the past six months or so. Then over the weekend I had a low-level background internal screaming going on, and while my first couple rounds of attempts at resolving it only helped a little, I was finally able to isolate the issue and fix what turned out to be a massive misalignment. I have not felt this aligned in years.
So thank you very, very much for writing this.
For the specific case of weighing yourself, could you create a scale that only gives the positive reward, not the negative one? Like, it only tells you your weight if it's lower than yesterday, or better yet if the trend in your weight is downward over the past week? Maybe it displays a cheerful message and plays a soothing sound when you weigh yourself, and it emails you later at random if you've been losing weight.
My fiance likes it, supposedly it's been very true to the original manga. It might just be better than the original dub.
At this point, I won't be confident that i've been successfully preserved until ultra high resolution electron micrographs of my brain are in Amazon's S3 storage, replicated across multiple regions. Any storage that doesn't have redundancy doesn't count as safe.
Is Austin on the list? I work at a not-evil tech startup called SchoolAdmin that does school admissions software for a mix of public/private/charter schools. We're not hiring devs right now, but that might possibly change since we have a product manager coming in October. The company is REALLY not evil; we've had three different people come down with mental or physical health issues, and the president's mantra has been 'your job is to get better' in every case.
I could possibly also offer a place to crash, I've got a futon, a study it could be moved to, and already have cats.
Body armor only protects against weak curses like Somnium that are standard first year hexes. Stupefy, which is castable by first years, is not blocked by armor.
He got Lucius to agree to proposed next steps which we don't know about and secured an alliance with House Malfoy.
Also, when he added the bit about exonerating House Malfoy, he showed Lucious one wording, then wrote down his original wording. He might have changed something important, and Lucius might not have noticed.
Any chance you can refactor your recipe to replace bananas with something that keeps better? Otherwise, I was pretty pleased with it.
I'd be interested to know how many people flunked out of that internship because they couldn't pick it up, and to what extent people were pre-selected for the internship based on some estimate of their ability to pick it up.