Hi, this is Wayne again with a topic “AI Experts: “What Have We Done?””.
Ai experts Advocate pause on high-level AI development. Over 1 000 experts, including Engineers, researchers and notable Tech luminaries, have signed an open letter asking for a six-month pause to development of AI, beyond the bounds already reached by gpt4. Their argument is that in the recent rush to develop AI systems, we have not had the necessary time as a global Society to fully consider the risks of developing human competitive AI intelligence. The letter further argues that if private entities are unwilling to restrict themselves voluntarily, governments should step in to regulate AI development. Some notable signatories include Elon Musk. Can I stop that for a second stop? What isn’t he like anti-government regulations? Could he be? Could he be less consistent in some way? Is that possible for him to be less consistent to to because, okay, to throw him a bone? In this scenario, he has been highly concerned about AI for a super long time and been calling for regulation for a super long time. Yes, yes, but he only calls for regulation when it doesn’t affect him. So that’s this regulation – oh okay, hold on okay! I’M gon na pause. My response to that, because I want to finish the notes, but I have a I I’m gon na jump on your side in a moment um.
This has led to crit uh, Elon Musk, Stephen Wozniak and the CEO of stability AI. This has led criticisms that the true motive behind some signatories is to slow down AI development so that their own AI Investments can catch up to the competition. Now we’re done the notes, I’m going to jump back on that, I’m very certain that this is specifically at least as far as Elon is concerned. Yes, trying to Target open AI because he’s really bad, because this will 100 only Target open AI right now. There’S all these articles about how like Google’s Bard, is literally trained on gpt4 outputs. There’S there was this website. I don’t remember what it was. I don’t think it was ran by openai, but it was like a collection of people’s conversations that they had had that they submit to the website.
So people can see all these different conversations that have been had and they were training it off of that website. They were trying to get Bard’s results more similar to the results that they were seen in that website right, and they got caught doing this an open AI. Like knows they did it at like they are absolutely the target of this also. I have no faith whatsoever that we, because we’re talking about the global Community, would be able to come to some conclusion in the next six months about how this is gon na go.
The global Community can’t come to a conclusion about anything like not murdering each other. All the time, we can’t agree that that’s gon na get no idea with this yeah also like, if, let’s say, uh, the government of a very large militarized company or country, sorry yeah same diff um if they wanted the benefits. No, it’s not me with the spicy. Take this week, if they wanted the benefits of some system, you think they’re gon na pause.
It no, of course not yeah. I don’t know I in like everything, goes perfect world like sure, let’s slow it down and figure out how we’re gon na do things, but like people that have if, if these exist yeah, if there are competitive systems with gpt4 right now that just aren’t public, for Whatever reason uh held by governments or other companies or whatever you think, they’re gon na stop, no, of course not even if you’re shocked, even if you told them to stop and they were like yeah, we stopped they’re, just gon na keep growing anyways like what. What is everyone who works? They’Re gon na do yeah twiddle their play, play ping pong for six months, like there’s, probably other stuff that they can do too.
This is another thing I thought of like. If this somehow does go through, you think open as just gon na completely sit on their hands they’ll just build like other supporting things yeah. So when they’re able to work on it again, they just ramp even further uh. I don’t know some of the uh development. That’S happened around gpt4 and the hook stuff is really crazy. My account doesn’t have access to plugins, so I haven’t been able to do it myself, but watching it.
I’Ve seen other people’s results watching it use the plugins to think has been pretty wild uh because, like I’ve, I’ve known about Wolfram Alpha for a long time, which is that, like math tool, I didn’t I haven’t used it since I was in university. I didn’t realize that it’s Error output is written in like human language. Oh, that’s cool that helps gpt4 a lot right, there’s also people that are already setting up tools making it so that, even without the support of open AI, they have like there’s a paper on this.
Someone wrote a paper. I was extremely interesting where they set up a system uh a series of tools that all conversed back and forth so that they were able to externally make the project that open AI was talking about about giving openai access to itself because they just linked multiple accounts. Together through One controller yeah, and then they gave the controller the task of telling the other accounts that they’re like what their job is in the queue and getting everything built, and it’s all happening it matter.
If you tell them to pause, they’re going to keep going, anyways, yeah um, and if it’s not like, if you put tons of watchdogs on open AI, I’m sure they’ll find some way to work on something open AI as a organization. As far as I can tell, which this could just be them trying to convince me in succeeding, I have no idea, as far as I can tell, they are somewhat concerned about the ethics of all this kind of stuff, but they’re marching forward anyways. That’S what people are going to do? Yep? We are actually going to destroy ourselves. Oh yeah. We all know that right, yeah, like for sure yep. I was talking to you about a thing that I don’t know if I want to uh say yet, because I don’t want to like claim it until I’m more sure but uh another project that I thought that’s been really cool is: do you know about llama yeah Uh, so it’s like locally ran people have set up uh like microphone arrays in their house and some someone posted a project where they made it so that they use Siri voice whatever through a MacBook, to write into a a like lamba input so that they could Converse with it back and forth through voice – and I know someone else – who’s been setting up a microphone array in their house so that they have like a Jarvis like that project is happening now.
Yeah you’re not going to stop this stuff, and especially when you get to the point where people can just ask chat, GPT to write whatever Plug-In or whatever hook. They need because you because, like the problem right now, is that these voice assistants or AI assistants, or whatever they can’t do anything. You have no reason to talk to them, but if any idiot with vocal cords can just say or or or hands or any way of inputting text can just say, uh, oh write a thing so that chat, GPT or llama or whatever can control my smart blinds.
Here’S the brand find them on my network or whatever it’ll actually do stuff, which is which is cool yeah. It’S interesting, yeah, good luck, everybody! I don’t know what what is your? What is your thoughts on the pause? You? Well, it’s just asinine yeah good luck with that. Yeah uh! It’S it’s like uh. How would they even? How do you think they would even try to enforce something? It’S like it’s. I think it’s like it’s like a ceasefire in like a in a in a traditional War yeah just because we’re not firing rounds doesn’t mean we’re going to stop manufacturing them, everyone’s just going to Super focus on Logistics and troop movement and intelligence, and all this other Type of stuff, so that they’re as ready as but think about uh uh, like peace in Civilization. We were talking about save earlier and yeah in the show.
What happens when you declare peace in Civilization? You both build up as much as you can, because it ends in 10. Turns like let’s go it’s coming. I don’t know uh atomic age Ace, I think, has it nailed down? What we really need is an AI trained entirely off of twitch chat. Sorry I just that amused me.
It was in the floor. Plane chat like that’s a way that they could work on it without making it better necessarily is something that’s been brought up is uh that making an AI that is like uh good in the sense of like good nature, bad natured, whatever is basically impossible due to All the different cultures in the world, because you’re gon na be like my morals, are good, but that’s not necessarily going to line up with someone somewhere else, so you almost need to make it so that those parts are like modular right. So, like, oh man, you know what I am. It gets really messy really. I am not gon na pick a polarizing issue to say like that, could be a module, okay, that you would unplug or replug, depending on the part of the world that the AI is operating in or political affiliation or whatever yeah exactly yeah. But but like is that the right route to go? I don’t know I’ve seen I’ve seen some some people big mad about AIS, spitting out like objectively factual stances on some very polarizing issues. The significant portions of the population do not understand well enough to know that. That’S actually just a way better answer than what they think it’s like, like big mad, so biased um, so it’s like they.
They could spend this six month six months, trying to tackle like that problem that doesn’t technically Advance its. They probably just won’t bother. I don’t know it: there’s yeah, it’s going to be interesting. Future is very uncertain. I I feel bad for people that are in, like you know, like late high school right now, because you’re in that stage, where you’re like okay, I’m leaving the like the nursery. The standard required education system, the the walk through either school or get a job or whatever, what career path? Do I put myself on when I have very low experience, because a lot of the jobs that are most at risk, our non-expert style positions, but then how do you get to an expert style position if you can’t gain job work? Experience so like, if everyone just replaces all the low-level jobs, how do you fill the positions of the high levels, so we become batteries? Tell me more Neo, oh my God.
We already are: oh man, um. Aren’T we all just training data .