Co-Founder and Director at https://ai-plans.com
Please get in touch if you’re:
- A skilled JS developer—
Interested in joining the Critique-a-Thon ( https://ai-plans.com/contest )
- Interested in being a judge for the Critique-a-Thon
Kabir Kumar
[Question] AI Safety orgs- what’s your biggest bottleneck right now?
Extremely important
This is an absurdly low bar, but yes, this should be done.
How can I help?
[Question] Do you want to make an AI Alignment song?
Yes, that’s what I’m referring to. As in, getting that enacted as a policy.
Perhaps a note on Pre-Requisites would be useful.
E.g. the level of math & comp sci that’s assumed.
Suggestion: try going through the topics to 50+ random strangers. Wildly useful for improving written work.
No Clickbait—Misalignment Database
Updated to 115.
Ah, sorry, here’s the link! https://docs.google.com/spreadsheets/d/1uXzWavy1mS0X-uQ21UPWHlAHjXFJoWWlN62EyKAoUmA/edit?usp=sharing
Thank you for pointing that out, also added it to the post!
Thank you! I’ll add those as well!
Thank you, I’ve labelled that as the form link now and added the DB link.
for sure. right now it’s just a google form and google sheets. would you be interested in taking charge of this?
Week 3: How hard is AI alignment?
Seems like something important to be aware of, even if they may disagree.
Yet, the top posts on LessWrong are pretty much always clickbaited, just in the LessWrong lingo.
The Curated Posts seem to be some of the worst cases of this:
I find that LessWrong folk in general are really, really, susceptible to deception and manipulation when it’s done in their language.
Hasn’t that happened?
Thank you, this is useful. Planning to use this for AI-Plans.
You might be interested in the Broad List of Vulnerabilities
My Clients, The Liars
And All The Shoggoths Merely Players
Acting Wholesomely
These are the most obvious examples. By ‘clickbait’, here I mean a title that’s more for drawing in readers than accurately communicating what the post is about. Doesn’t mean it can’t be accurate too—after all, MrBeast rarely lies in his video titles—but it means that instead of choosing the title that is most accurate, they chose the most eye catching and baiting title out of the pool of accurate/semi-accurate titles.
Ok, so are these not clickbait then?
”Stop This Train, Win a Lamborghini”“$1 vs $250,000,000 Private Island!”
“$1 vs $100,000,000 House!”
“I Hunted 100 People!”
“Press This Button To Win $100,000!”
there is an issue with surface level insights being unfaily weighted, but this is solvable, imo. especially with youtube, which can see which commenters have watched the full video.