Artificially Intelligent

Any mimicry distinguishable from the original is insufficiently advanced.

  • Interpolate Claims (Un)charitably

    | 438 words

    The principle of charity requires interpreting the speaker’s claims in the strongest possible light. This principle is ambiguous. If I say “money can’t buy happiness,” the truest version is vacuous: “there is no way to convert money directly into happiness”. More nuanced notions of strength are context-dependent; “strongest possible light”...

  • The First Sample Gives the Most Information

    | 453 words

    I originally heard this point made by Ben Pace in Episode 126 of the Bayesian Conspiracy Podcast. Ben claimed that he learned this from the book How to Measure Anything, but I think I identified the relevant section, and this point wasn’t made explicitly. Suppose that I came up to...

  • Defusing AGI Danger

    | 2807 words

    Crossposted from the AI Alignment Forum. May contain more technical jargon than usual. This represents thinking about AGI safety done under mentorship by Evan Hubinger. Thanks also to Buck Shlegeris, Noa Nabeshima, Thomas Kwa, Sydney Von Arx and Jack Ryan for helpful discussion and comments. tl;dr A common perspective to...

  • Chain Breaking

    | 3126 words

    This is a rationality technique I’ve been experimenting with. Thank you to Jack Ryan, Thomas Kwa, Sydney Von Arx, Noa Nabeshima, and Kyle Scott for helping me refine the method. Algorithm Pick something that has happened before that you would prefer not to happen again. Examples include: not exercising, not...

  • CFAR Retrospective

    | 1434 words

    I recently read CFAR Workshop in Hindsight, which had the following tl;dr: I believe it changed me in subtle ways and improved several skills (mostly “meta skills”) marginally. This differs quite a bit from the somewhat overblown expectations I had before the workshop, but I’m still quite certain the workshop...

  • My Routine

    | 1121 words

    Recently, I’ve been about as productive as I’ve ever been. One reason for this is my current routine, which can roughly be described as “systematize literally everything”. I currently work from home. Here’s a snapshot of what my routine currently looks like. Monday-Saturday ~0800: wake up 0800-0900: eat breakfast, respond...

  • Be Responsible

    | 570 words

    In high school, seniors go on a seven-day class-wide camping trip. When it came my turn, I was placed into a group of 28, in which we formed groups of four. Each of these groups was responsible for one night of dinner. My group of four decided that we wanted...

  • A Math Student's Guide to Options

    | 2709 words

    Disclaimer: This represents slightly more than everything I knew about options when I put my entire net worth into SPY puts in March.1 I learned most of this in about 30 minutes of reading. This introduction should be accessible to someone that knows statistics and calculus, but little finance. The...

  • Be Goal-directed

    | 546 words

    A question that I’ve recently grown fond of is “what’s your goal?”. Usually, this question arises when someone is describing course of action and wondering if it’s good or bad. Sometimes, during this process, people get trapped in this mindset where they think the course of action is either good...

  • Does SGD Produce Deceptive Alignment?

    | 4905 words

    Cross posted from the AI Alignment Forum. Deceptive alignment was first introduced in Risks from Learned Optimization, which contained initial versions of the arguments discussed here. Additional arguments were discovered in this episode of the AI Alignment Podcast and in conversation with Evan Hubinger. Very little of this content is...