Home
Creators
Search Recent Random
Posts
Search DMs Popular Hash Lookup Tags Random
Importer
Import FAQ
Register Favorites Keys Review DMs Logout
Community
Telegram ThePornDude
Home Creators Posts Import Register Favorites Logout

Early Access: Why Would AI Want to do Bad Things? Instrumental Convergence (Patreon)

Published:
2018-03-23 16:14:45
Imported:
2026-06
Tags:
haven't archived this post yet. have a subscription? use the importer!

Content

How can we predict that AGI with unknown goals would behave badly by default?

Files

Previews only

Why Would AI Want to do Bad Things? Instrumental Convergence

How can we predict that AGI with unknown goals would behave badly by default? The Orthogonality Thesis video: https://www.youtube.com/watch?v=hEUO6pjwFOo Instrumental Convergence: https://arbital.com/p/instrumental_convergence/ Omohundro 2008, Basic AI Drives: https://selfawaresystems.files.wordpress.com/2008/01/ai_drives_final.pdf With thanks to my excellent Patrons at https://www.patreon.com/robertskmiles: Jason Hise Steef Jason Strack Chad Jones Stefan Skiles Jordan Medina Manuel Weichselbaum 1RV34 Scott Worley JJ Hepboin Alex Flint James McCuen Richárd Nagyfi Ville Ahlgren Alec Johnson Simon Strandgaard Joshua Richardson Jonatan R Michael Greve The Guru Of Vision Fabrizio Pisani Alexander Hartvig Nielsen Volodymyr David Tjäder Paul Mason Ben Scanlon Julius Brash Mike Bird Tom O'Connor Gunnar Guðvarðarson Shevis Johnson Erik de Bruijn Robin Green Alexei Vasilkov Maksym Taran Laura Olds Jon Halliday Robert Werner Paul Hobbs Jeroen De Dauw Konsta William Hendley DGJono robertvanduursen Scott Stevens Michael Ore Dmitri Afanasjev Brian Sandberg Einar Ueland Marcel Ward Andrew Weir Taylor Smith Ben Archer Scott McCarthy Kabs Kabs Phil Tendayi Mawushe Gabriel Behm Anne Kohlbrenner Jake Fish Bjorn Nyblad Jussi Männistö Mr Fantastic Matanya Loewenthal Wr4thon Dave Tapley Archy de Berker Kevin Vincent Sanders Marc Pauly Andy Kobre Brian Gillespie Martin Wind Peggy Youell Poker Chen Kees Darko Sperac Paul Moffat Noel Kocheril Jelle Langen Lars Scholz

Comments

loopuleasa

Let me contribute more to the terminal goal of "Educating the world", by fulfilling the instrumental goal of making your channel better through constructive feedback. 1. The first visual impression Get better thumbnails. A face helps a lot, statistically, so you are doing that right, but the font is too much I feel. Black background is good as it will attract nerds that are used to Brady video thumbnails. Try to replicate other edu YT channels, since our neural nets are already focused on identifying what we like or not, in our bubble. The problem is, we are smart enough to identify what is quality and what is not, and you channel is still budding. Make a good first impression in order to scale. The quality is growing. 2. The face video format hypothesis. Since your face is attractive, and you have an endearing attitude, the full-on face format will work fine, but don't neglect for a moment the visualizations that appear. I love what you are doing with the big-word-terms-on-black-background-fade-over-your-face things. Similar to Vsauce, but more readable. Make that a theme. Also, good use of the negative space around your face for other visualizations. Please fill that up, and strike a balance. Right now, I feel you need 20% more visualizations, and of quality too. Less memes, that is, unless they fit really well. 3. Self-improvement for a youtuber. I recommend looking on Veritasium and maybe reading his studies. He had a thesis on making science videos for a PhD or something, he has some useful findings you can use. I will mention some key findings now that come to mind, but dig deeper yourself for the channels sake, you have the brain power to optimize on that. Bonus, there is a whole youtube section for creators, called YouTube creators or something. Look at that when that becomes a priority. 4. Emotion. Veritasium mentioned that people's attention goes through emotion, instead of rational. Your analogies and stories are really good. I like how you weave them. Just don't forget that no matter how nerdy we are, we are still emotional bundles, and utilize that to spread the good. 5. Humor is good. Keep it up. Your channel will probably die without some pop to spice things up. Your flavor of sarcasm is quite tasty. No homo. 6. Always start with the misconceptions Veritasium also had a video on that. If you want people to learn something in particular, let them answer it in their head. They will answer with a misconception. Then you can clear it up. Don't give the answer straight away, let our brain fight for it. Telling you all this because I am also interested in a channel, since I obviously have usefully interesting things to say, but less time to spare. Wanted something of the style of exurb1a, which I recommend tremendously to watch and emulate some concepts, you might feel inspired since it's all over the place and it's 5 star material. If I see a trend, I will also double my patron bonus. FInally, make the YT channel a second priority, and focus on what makes you credible in the first place: AI Safety research. Keep us posted with what happens in the field. Lup.

loopuleasa

Oh, you have a second channel. Yus. Btw, you should mention and cross-ref your second channel videos during or at the end of your main channel videos. This helps a lot at getting people emotionally invested in your journey. Have fun.

The Nanomatic
>>11061710

Excellent and well thought out feedback. Even I appreciate it. I think all of the video topics covered thus far are interesting, to say the least. To note, nothing of video content in particular I feel is lacking, and is already being done righteously. Furthermore, I adore this style of feedback. I will have to use it myself sometime.