Це відео не доступне.
Перепрошуємо.
The Lie of AI Alignment
Вставка
- Опубліковано 15 сер 2024
- The idea of "alignment" has been a key issue in AI for quite some time now. Let's break down the reasons why AI "alignment" as it's been sold to us is a lie, and definitely won't benefit you or me.
0:00 Intro
3:12 Will Monopolisation of AI by Large Organisations Make it Safer?
8:08 What is Alignment?
19:15 Conclusion?
twitter: x.com/real_cha...
discord: discord.gg/tSb6QpxcUS
"The best way to stop a bad guy with a vial of smallpox is a good guy with a vial of smallpox"
completely missed the point bud.
Great video btw
@Chadeus Disease control facilities keep small pox and other pathogens available for usage, as with vaccines, and study.
@@Chadeus yep, bad guys don't want to get smallpox either, and there has definitely never been any examples of any bad guys throughout history that were more than happy to put themselves and their loved ones in harms way in order to achieve their goals. yep... yep.
AI is a Pandora's box that bad actors are just dieing to get their hands on.
Look up extinction cults...
OK, going to watch the video, but what I guess "safe" means for the people involved is simply "it does not threaten our positions of power" simple as that, and of course "it aligns when AI helps us to further cement our positions of power". Its sad, but isn't that what "they" always tried, for like the last 2000 years or so?
Where does proper ideals win over the money game? Where's the evidence that's ever going to happen? Our hospitals can't even get past the money driven issues. How much of the AI fear is real versus Hollywood. What evidence are they basing their fears on? I'm a programmer btw. There's a right way but we don't do that because there's a game to be played. Ideals versus our half-assed systems. Good luck to us all.
Well there is a list of incidents of specification gaming about 5 pages long any given one of the failure modes that occurred there being present in a general intelligence would really fuck us over
good video about OpenAI, but what does this have to do with AI alignment theory?
here too please, thank you 🙋♂
09:59 - where did you get this video ?
Something is misunderstood here, because it’s questionable how much or how little leveraged risks AI can truly carry. Compare it with the dangerous technologies of before, the most common example being the nukes. AI alignments researches and talking heads have basically accepted as an axiom that AI is a nuke in disguise and we should treat it as such (with monopolisation, restrictions, ban on testing and so on, and so forth). So I really applaud your criticism of this inflexible and sometimes insidious way of thinking.
At the same time, to make a case that organic spread of technology is the best viable alternative concerning AI risks requires more evidence that AI is akin to or perhaps safer the deployment of internet, as an example of fast pace organic development.
When many shareholders are often the same small collection of hedge funds...
The more open source AI is the better.
I don't trust Sam but I trust North Korea less
your dilemma is simple, which of the two wish to inflict full spectrum dominance
Sam harris is jevzzzzz propagandist and north korea minds their own business
I'm on a streak of watching people making comments that they should have thought twice about. you seem to have a good head on your shoulders but did you really mean to suggest that mentally ill people pose the same threat as terrorists if given access to technology?
are you saying terrorists aren't mentally ill? you should maybe re-think that as well