transitivebs

Ran into a really interesting illusion working on a parody video with Opus 5.5 today.

The experience of having Claude tell me I'm wrong (and it being right) and then learning about the McGurk effect were both interesting enough that I had to create a lil explainer + writeup on it.

"curiosity driven queries" – love this phrase, i'm def going to start using it :)

guide to how it was created: https://felixrieseberg.com/made-by-mechanical-means/

he wrote in-depth about how it was created https://felixrieseberg.com/made-by-mechanical-means/

an in-depth post about how he made it: https://felixrieseberg.com/made-by-mechanical-means/

extremely fascinating details in here

Took some time to write down my predictions for AI cybersecurity.

TL;DR AI is turning cybersecurity into ~drone~ swarm warfare, and this will lead to increasingly consequential cyberattacks over the next 18 months.

openai will take expert responses like this and improve the next set of papers

it won't be long before there's no more low hanging fruit like this to complain about, and the writing / explanations of the results are superhuman as well

separately, i really liked the author's denial-of-service analogy. super useful practical framing

somebody bought dot.com to confuse openai dots.com users by redirecting to Grok bot lmao

this is exactly why i worked on webtorrent ~13 years ago https://github.com/webtorrent/webtorrent

yeah, you're right bro. sorry for the glib response; you were engaging in good faith, i just didn't really take it that way

not anywhere close to my 50s, and this isn't "bad-faith" marketing game; everything i do is 100% free and open source and just genuinely trying to help advance convos around AI in a positive direction without caring about making money like most people

anyhow, sorry for the rude remark

curious who would you like to see added on here?

seems like the assessment was pretty accurate based on the demonstrated reasoning in this response..

definitely; what you're referring to is known as extreme concentration of power is an aspect of more doom-type outcomes. it's not necessarily as existential as the extinction of all humanity, but it is accounted for in the model

i go into this more in this project: https://cultural-alignment.com/concepts/power-concentration

glad you found it helpful :) and yes, i find ed zitron super painful to listen to in general

there are some more people here in the larger view of X users: https://www.doom-or-bloom.com/users

all of the pages include the public sources they're based on and a clear disclaimer. this is also spelled out in the methodology section. many of these folks are using direct quotes to answer the questions where available as well

that being said, i agree, the simulated personas could be better flagged

appreciate the feedback

hmmm the system's def not perfect. hopefully the process was still useful?

thanks for your feedback; if you publish your results, i'll look into it and try to figure out where things went wrong

here's the results of your 2 answers posted into this tool btw. your views are remarkably close to ed zitron's (no judgement; just an observation): https://www.doom-or-bloom.com/public/assessments/8c615299-46...

1. this is just a useful marketing framing. the actual assessment uses a much higher dimensional space to model your worldview in a more meaningful, nuanced way

2. by expected bias, i meant that your answer didn't seem to actually look past the name at the actual content of the project or else you would've realized there are plenty of people modeled on here who share similar views as your own

3. i agree this type of knee-jerk reaction to the name & featured framing is reasonable. always tradeoffs when trying to get an idea to catch hold

good call; will update the sign-in to be less prominent and a disclaimer. appreciate the feedback

which is why it's so important for people to answer this question.. your views are 100% supported by this project if you leave your expected bias at the door ;)

yeah, moloch will come for us all in the end

i don't have net worth data in there, but you can sort by X follower counts on this more detailed page: https://www.doom-or-bloom.com/users

i really spent a lot of time looking through results and trying to course-correct and fix the model where i found obvious issues, but it's definitely not perfect

if you find any particularly egregious results, they're likely due to not having good enough public sources for that user's viewpoints. def let me know and i'll try to improve them

thanks

i agree, but if you view the sources his answers are grounded in, he seems to have updated to being more moderate in light of recent progress. still def on the lower side of the predicted impact spectrum, but it seems he's taking the impacts more seriously now

that's why it's so important to show the sources used to derive these positions. like https://garymarcus.substack.com/p/why-my-pdoom-has-risen-dra...