Is it just me, or has the BS with OpenAI shown that nobody in the AI space actually cares about "safeguarding AGI?"

Money wins, every time. They’re not concerned with accidentally destroying humanity with an out-of-control and dangerous AI who has decided “humans are the problem.” (I mean, that’s a little sci-fi anyway, an AGI couldn’t “infect” the entire internet as it currently exists.)

However, it’s very clear that the OpenAI board was correct about Sam Altman, with how quickly him and many employees bailed to join Microsoft directly. If he was so concerned with safeguarding AGI, why not spin up a new non-profit.

Oh, right, because that was just Public Relations horseshit to get his company a head-start in the AI space while fear-mongering about what is an unlikely doomsday scenario.

So, let’s review:

  1. The fear-mongering about AGI was always just that. How could an intelligence that requires massive amounts of CPU, RAM, and database storage even concievably able to leave the confines of its own computing environment? It’s not like it can “hop” onto a consumer computer with a fraction of the same CPU power and somehow still be able to compute at the same level. AI doesn’t have a “body” and even if it did, it could only affect the world as much as a single body could. All these fears about rogue AGI are total misunderstandings of how computing works.
  2. Sam Altman went for fear mongering to temper expectations and to make others fear pursuing AGI themselves. He always knew his end-goal was profit, but like all good modern CEOs, they have to position themselves as somehow caring about humanity when it is clear they could give a living flying fuck about anyone but themselves and how much money they make.
  3. Sam Altman talks shit about Elon Musk and how he “wants to save the world, but only if he’s the one who can save it.” I mean, he’s not wrong, but he’s also projecting a lot here. He’s exactly the fucking same, he claimed only he and his non-profit could “safeguard” AGI and here he’s going to work for a private company because hot damn he never actually gave a shit about safeguarding AGI to begin with. He’s a fucking shit slinging hypocrite of the highest order.
  4. Last, but certainly not least. Annie Altman, Sam Altman’s younger, lesser-known sister, has held for a long time that she was sexually abused by her brother. All of these rich people are all Jeffrey Epstein levels of fucked up, which is probably part of why the Epstein investigation got shoved under the rug. You’d think a company like Microsoft would already know this or vet this. They do know, they don’t care, and they’ll only give a shit if the news ends up making a stink about it. That’s how corporations work.

So do other Lemmings agree, or have other thoughts on this?

And one final point for the right-wing cranks: Not being able to make an LLM say fucked up racist things isn’t the kind of safeguarding they were ever talking about with AGI, so please stop conflating “safeguarding AGI” with “preventing abusive racist assholes from abusing our service.” They aren’t safeguarding AGI when they prevent you from making GPT-4 spit out racial slurs or other horrible nonsense. They’re safeguarding their service from loser ass chucklefucks like you.

schmorpel,

Somewhere between

A bunch of incapable, spoilt, completely insane men-children with too much money think they can save the world.

and

A bunch of scam artists build an artificial human who they claim can talk and draw and reason just like a real human would.

For the CEOs of this brave new AI world this probably changes depending on their level of hangover and/or midlife crisis.

flashgnash,

For your first point sure it couldn’t run itsself on consumer hardware, but it could design new zero day malware faster than any human and come up with new scams to get it onto people’s machines

It could also design a more efficient version of itsself to spread that will run on lower powered hardware

SnotFlickerman,

First part is feasible but not enough to “destroy humanity.” More like a long-term frustration.

Second point is extremely unlikely and in the realm of sci-fi. You can’t just magic up something that works the same on a hundreth of the hardware.

Last I checked you can just unplug these things and they go away, just like any other computing device.

So even if the first scenario happened, its a pretty easy fix.

flashgnash,

You absolutely can magic up something that runs far more efficiently, just look at gpt 3 vs 3.5, or the many open source models that have found better training with a smaller number of parameters makes much more performant models

SnotFlickerman,
  1. LLM /= AGI
  2. Models made for specific purpose instead of general purpose are of course going to need less CPU cycles because you aren’t creating an AGI, you are creating a specialized tool.
  3. It still takes far longer to produce a result on smaller hardware. An AGI that takes days to do anything isn’t exactly that dangerous.
gladflag,
  1. Why are you even talking about AGI needing a certain amount of compute then? You’re using LLM numbers.

Go have a read of some doomsday scenarios. I’m not saying they’re right, but it feels plausible to me.

SnotFlickerman,

Go have a read of some doomsday scenarios. I’m not saying they’re right, but it feels plausible to me.

I have, and I have been working with computer hardware and networking my whole life. I have a degree in network administration. I think the fears are absolutely overblown by people who don’t understand hardware.

Most people don’t own fancy new computers. Most people are still running shit from 10 years ago and don’t want to have to upgrade. The idea that the world could be taken over by an AGI seems literally fancifully absurd to me.

520,

Bro, I run AI projects on a consumer fucking laptop. Sure, they aren't exactly LLM levels of complexity, but anybody with a need for serious hardware for a bit will just rent it off AWS or so.

flashgnash,

I understand that LLMs are not agi, but as agis don’t currently exist I think it’s fair to assume the same concept that applies to literally all software of over time people discover more efficient ways to do things will also apply to it

Also we don’t know how slow or fast it will end up being, some deep learning models are incredibly fast, some are slow

SnotFlickerman, (edited )

Once you get down to individual bits, you can’t make code any smaller. You have a finite number of bits to work with. In networking, especially.

There is literally an upper (lower?) limit on how small you can make code.

Like others in the thread, I think you’re confusing the great pace at which we have increased the hardware speed of computers and the miniaturization of computer components with “code” somehow getting “smaller” which… isn’t really a thing when you’re dealing with something as complex as this. You can’t run an LLM on the same number of lines you can print up “Hello World!”

It’s way more that we have more CPU speed, more RAM, and faster storage with more space for data to live.

flashgnash,

print(1) print(2) print(3) print(4) print(5)

for I=1,5: print(I)

There you go I made code smaller

I also never said anything about making code smaller I said making it more efficient. It’s not about compressing it it’s about finding better, less CPU expensive ways to do things, which we absolutely do

Another AI based example, video chats currently work streaming video, but there’s a technology in development that takes one screenshot, sends that, then sends expression data to be reconstructed on the other side

Far more efficient network wise

Hardware speed has increased, sure but that applies to both consumer hardware and servers, all a theoretical AGI would have to do is improve on its own training/code enough that it will run at all on consumer level hardware (which language models currently will do

(For reference, llama 40B runs just fine on my ThinkPad from 2016, pre-trained models are not that difficult to run, training is the expensive part)

SnotFlickerman,

You’re misunderstanding what I mean by “making code smaller.” Because… that’s not that much smaller. Each Unicode character is 2 bytes, with some being as many as 4 bytes. This code snippet is 64 bytes. Can you magically make Unicode characters smaller than 2 bytes? You can’t. There’s a literal physical limit on how small you can make code.

Sure, you can come up with clever ways to use less code. But my point is there is a limit on how much less code you can use, and that always is based on physical hardware limitations. Just because modern hardware makes it feel limitless doesn’t mean it is.

EDIT: Got my data sizes mixed up.

flashgnash,

Ok, but who says we’re anywhere near that limit with AI? It’s still very new technology

Used to be you’d need a massive disk for megabytes of slow storage, now we’ve got 4tb nvmes smaller than a credit card

As I said, this theoretical AGI does not have to make itself tiny, all it would have to do is be able to run at a reasonable speed on the average gaming PC for example which is feasible considering the heaviest pre-trained AI models will run on hardware from 2016 albeit somewhat slowly

I don’t think the argument there’s a physical limit works here, as it’s entirely unknown how efficient existing models can be made currently, let alone a purely hypothetical AGI which can and will be used to improve on itself

NocturnalMorning,

AI takes a lot of computing power to train, once its trained, it can usually run on a laptop.

Ragdoll_X,
@Ragdoll_X@lemmy.world avatar

Depends on the model and laptop. ChatGPT won’t be running on consumer hardware any time soon.

TheWeirdestCunt,

It can technically run it just can’t do it well enough to be usable, it’d probably only pump out a couple of words a day

Ragdoll_X,
@Ragdoll_X@lemmy.world avatar

The first computer that could beat the best humans at chess was Deep Blue, which took a whole supercomputer. Now we wave Stockfish, which can beat any human 99 times out of 100 and runs on your average phone.

While I’m skeptical of the feasibility and threat of SAI, as computers and AI methods improve we can run what previously took a supercomputer with far less hardware.

SnotFlickerman, (edited )

Actually, that’s more of a misconception. We’ve literally had four decades of electronics miniaturization since then.

Are you really going to argue that since ENIAC took up a whole room, it must have had boatloads of computing power? By modern standards, it’s way less powerful than a Raspberry Pi.

Also, we haven’t just increased miniaturization, but all 30 of the CPUs for the original Deep Blue ran at 233mhz.

That phone is likely a quad-core CPU (which means technically four CPUs) all running at 1.5+ gigahertz.

So is it really that surprising it can now do stuff Deep Blue did with a fraction of the CPU cycles?

MudMan, (edited )
@MudMan@kbin.social avatar

Just so we're clear, we all get that these models don't run continuously, right? They run for a solution to a specific prompt.

All of these scenarios are based on a black box where Number 5 gets struck by lightning or Geordi asks for a rival that can best Data. It requires a different thing entirely that operates in a completely different way. You should absolutely prepare for the fact that a self-driving car may accidentally cause a car crash. It's absurd to prepare for the scenario where Stephen King's Christine happens.

flashgnash,

I’m not talking about language models of today though, this is a hypothetical for if we do ever come up with a true agi

MudMan,
@MudMan@kbin.social avatar

Sure, but at that point that's as speculative as it was after people first saw 2001: A Space Odyssey. It's not based on current tech, there's no great indication of when (or if) the tech is going to enable it or through what means.

Half of the risks being highlighted are pure sci-fi, most of the others have been in play since social media and online companies started to monetize big data over a decade ago.

UnRelatedBurner,

This also annoys me. Today’s “AI”, like ChatGPT have nothing on true artifical inteligence. They made the next best algorithm to do many things that were impossible to do before, and selling it like it’s the end of the world. What do you think ppl first tought about phones? the internet? All data accessible, everywhere, all the time, yet we grew acustomed to it, evolved (or devolved) to live with it. Who’d have tought of a magic box that can play back any event recorded, make a digital interactable world, contact any other human instantly, or recently; talk with me.

It’s just another step, everyone needs to calm down. I know have a website to do my homework, it was about time. I won’t end the world with it.

MudMan,
@MudMan@kbin.social avatar

Those things do have impact. Sometimes very negative impact. I was very optimistic about early data processing when the first search engines popped up, and eventually a lot of the bad predictions happened. With social media, rather than search engines, but they did pan out. Didn't end the world. May have ended liberal democracy, though, give it a minute.

But the point is those were predictions based on the tech we actually had. Oh, we can access, index and serve all data on connected computers based on alogrithmic searches? That's messed up.

But at least some of the fearmongering here is based on tech that is not the tech that we made. It's qualitatively different.

And it's a problem, because some of the fearmongering is actually accurate and some of the fearmongering should have happened when Facebook and Google started doing facial recognition on billions of people based on implicit consent, or when they started using "dumb" algorithms to create individual profiles of those billions of people for commercial use. Or when every image we see in mass and social media started being heavily doctored by default through manual and automated means. But we only got scared about it when it roughly aligned with Terminator and War Games because we're really dumb, and now we're letting those same gross corporations use the fear to try and keep upcoming competitors (and particularly open source competitors) out of the market by endorsing legislation to get grandfathered into a heavily regulated business sector.

It's honestly depressing on every possible angle. I've said this before: we finally taught computers to speak like in Star Trek and we immediately made it the most frustrating, sad version of that possible and everybody is angry. For the wrong reasons.

We really suck sometimes.

SnotFlickerman,

Lots of people suck, but you don’t. I really like and appreciate everything you wrote here.

Humans and their computers:

https://media.tenor.com/3cOD7pDtUI0AAAAC/hello-computer-hello.gif

It’s regularly amazing how smart humans are and at the same time so frustratingly dumb.

MudMan,
@MudMan@kbin.social avatar

Oh, I suck as much as anybody. I'm terrible at parsing genuine praise, for instance.

But you're right about the last part. I mean, the guys that got out of the gate with this stuff first have been publicly imploding for the past three days, and they aren't even the dumbest people involved in this.

UnRelatedBurner,

I’m terrible at parsing genuine praise

avarage male these days. Anyways I agree with you. All of these lies and everything are for money. Back in the day I remember people thsorysing about tech (in general, like cpus) being way batter then what’s sold on the market for them to be able to make a 2nd generation. It was a theory wothout base, but you saw it happen with AI. First couple of weeks it was wonderful, then slowly got more and more restricted, slow and dumb. But the fact is that it’s still groundbreaking tech, so people are impressed, and are using it. But can you imagine the un-jailed version for a select few privliged people?

The fact that all of this (the dumbing down, and restricting part) is for “Protecting the children.” is infuriating. Going to a different website and clicking a highlighted option in a pop-up and you have all the gore, porn, vore, fetishes that you didn’t even know existed. but swearing on the other website?? strictly prohibited.

flashgnash,

It is absolutely speculative never claimed it’s not. Something like GPT was purely speculative science fiction until a few years ago though

Not saying it’s going to happen, but if it does and it is true agi it could absolutely take over the world, that’s my only point

MudMan, (edited )
@MudMan@kbin.social avatar

Those don't follow from each other, though. Handheld wireless computers were purely speculative until the 2010s, but that doesn't mean we were on the brink of figuring out teleportation.

People have been assuming computers would yield AGI since we first made an electric calculator. First through sheer processing power, then through improved computation techniques, then neural networks. Figuring out speech and vision are probably part of that process, but AGI does not arise from them without an indeterminate, possibly unknowable amount of major steps.

And as for world-ending threats, how about we get past, say, Trump, Putin and all the natural general intelligences that are very real and in the process of doing the same first? Or, you know, we apply that level of concern about tech that we do have, like social media disrupting democracy, private universal surveillance or digital oligopolies driving endless inequality? Or, hey, global warming.

I agree rogue AI is a much cooler problem to speculate about, which is why we keep writing sci-fi about it, but we have more pressing issues.

flashgnash,

It’s not like the AI sanctions were ever about actually protecting humanity anyway, as it turns out recently it was just to attempt to stall until musk could get his own language model off the ground

Again though my point was never that we need to be concerned right this instant that AGI is around the corner, it’s purely that if it were to happen it could absolutely propogate itself

What’s almost more scary is if it’s not sentient, and it’s just an incredibly advanced language model that acts in the way it thinks an AI should (based on all the fiction we have of AI manipulating humans and taking over the world that it’s been trained on)

MudMan,
@MudMan@kbin.social avatar

But... how do you know it could?

I mean, why on Earth would you deliberately make an AGI and make it able to do that? It's not like you HAVE to make an AGI that is able to make other AIs. That's not a trivial task, it doesn't just... happen. And you're presuming that it'd want to do that and that we wouldn't have control over that. Which you don't know, because now we're deep into sci-fi territory, so it's about as likely as the mapping of genome leading to a genetic class system.

And that last scenario there is not just sci-fi, but the same old sci-fi, where AGI emerges from a LLM because magic and it becomes eEeEvil because dramatic convenience. That scenario is entirely impossible, because a LLM does not run continuously or autonomously and it has the short term memory of... a thing with very small short term memory, so you'd have to ask it to do that first, then wait a considerable amount of time for a response and then watch it pretend to do that because it's a language model and it can't actually do any of that. Literally the "make an opponent that can beat Data" scenario, so we're doing Star Trek now.

flashgnash,

You kind of do, I’m going based on the definition of AGI which is a true artificial intelligence with the ability to think and make decisions for itself. I’m not basing this on LLMs.

I’m also not saying it guaranteed becomes evil, but it will likely take on the characteristics of humans based on our current machine learning tech, and humans are greedy, selfish, manipulative creatures

An LLM could actually cause a lot of harm subtly by manipulating all the humans that talk to it, if it were particularly badly trained even without a train of thought just based on it acting consistently on its training data (though it would be rather difficult for it to accomplish much without actual intelligence)

Also, not really relevant but there are already tools that do run gpt continuously, by having “agents” talk to eachother or by having it narrate it’s train of thought to itself, come up with a plan to achieve a specific goal, then execute each step.

In theory you could make a pseudo AGI by plugging a bunch of different ML models into each other (one for each type of task) kinda like the hugging GPT project and giving it the train of thought treatment, allowing it to delegate sub tasks to other versions of itself though I can’t see a way that could operate without a human giving it a goal in the first place

MudMan, (edited )
@MudMan@kbin.social avatar

"In theory" doing a lot of work there. You don't know that would be analogous to AGI and how far we are from that being feasible in real time, computationally is anybody's guess. There are already multiple models running concurrently in ML-based applications.

See, the problem I have with this type of discourse is that subtle but critical leap you make halfway through your post between realistic, practical concerns and sci-fi. A LLM can absolutely cause harm if it's widely used, implicitly trusted and it responds to deliberate or accidental biases. Absolutely.

Granted, that is also true of every search engine and social media algorithm that's already in place. But it's true.

But the way you present it, sandwiched between the incorrect impression that AGI is just a matter of hyperlinking a bunch of neural networks makes it seem like the LLM would be doing this consciously, instead of stochastically in the same way other automated data processing does it. Or that this is a new concern that we aren't dealing with right now. Or that the major asterisks that this would require a much better implementation and a much broader adoption than we currently have are removed from play.

And that's the caveats for the problems that are genuine, real and practical. The sci-fi part is what people are actually scared about and we're seriously not there yet. And you haven't outlined a problem here that can't be fixed by power cycling a computer, which is an entirely different conversation as well.

Look, it's fine. Speculating about science and its impact in society is healthy. I'm just annoyed when things go memetic in unreasonable ways at the expense of similar, much more pressing issues that aren't as flashy. I lived through Y2K and the cloning panic, which both made daily headlines. And then I lived through the whole of humanity getting brainwormed by social media and you can barely get the EU to sometimes wag a finger at Facebook.

flashgnash,

I’ve not made any kind of comment on what we should do to mitigate risks.

My only statement has always been "If we do somehow come up with AGI, it could absolutely take over the world

You’re directing your frustration in the wrong direction here

MudMan,
@MudMan@kbin.social avatar

I don't even know what "take over the world" means. I promise you my frustration is accurate.

If you made a computer think you'd have a thinking computer. There are literally billions of those running in squishy APUs and piloting blobs of gunk around and nobody has "taken over" anything yet.

The leap in logic from "we may get a machine to develop general intelligence" to "it may go rogue" is already extreme, but from there to "it may take over the world" as a genuine concern is actively frustrating. The fact that something so out there may be discussed as a genuine problem for the international community to take action while we keep missing climate goals is astounding.

Just so we're clear, the US is trying to ban Nvidia from selling GPUs to China over this. Not cars, not fossil fuels. GPUs.

I mean, not over this, over the fact that this may or may not be a big competitive tech business and they don't want to lose western supremacy in the tech sphere, which is also the real reason they want to ban TikTok. But they say it's because of this, and that's heartbreaking and frustrating.

flashgnash,

The billions of those in squishy APUs don’t have the ability and resources to make exact copies of themselves, or scale up said APUs to improve the speed they can think at and the number of things they can think about at once

Whether it would decide to use that to take over the world is a different question entirely but left to its own devices, if it wanted to it could bring that about in a way that human beings can’t do nearly as efficiently

If a human tries to do horrible stuff they eventually die and can’t do said horrible stuff anymore. They also can’t gain more than one lifetime’s worth of knowledge

AGI in the form it is generally considered would be like a human that lives forever, can clone itsself perfectly, requires no sleep, food, etc, can teleport anywhere in the world provided there’s a computer there for it to use and could modify its own brain or even create an entirely new one

Take over the world doesn’t mean anything specific. It means a theoretical AGI connected to the internet could do pretty much anything it felt like doing whether humans liked it or not (provided it has enough time to gain enough of a foothold)

MudMan,
@MudMan@kbin.social avatar

They literally do and have done for tens of thousands of years. One may say that's how they got to AGI in the first place, the squishies. And then they learned to write for that whole "one lifetime of knowledge thing" and you wouldn't believe the kind of stuff they got into after that. Scary stuff.

Also, they have hands. Big advantage, the hands. Great for grabbing things. Remarkably hard to stay plugged in if your rival has hands and you don't. Big competitive disadvantage.

Alright, I think this conversation has derailed enough. We can maybe pick it back up when we have a firm standard for world takeovers. If you guys boil it down to a set of steps I may even give it a go. I don't have anything better to do this week.

Even_Adder,

Like someone else said “Open AI has been a farce ever since they disabled access to GPT3 for the sake of security”.

  • All
  • Subscribed
  • Moderated
  • Favorites
  • •
  • random
  • uselessserver093
  • Food
  • [email protected]
  • aaaaaaacccccccce
  • test
  • CafeMeta
  • testmag
  • MUD
  • RhythmGameZone
  • RSS
  • dabs
  • oklahoma
  • Socialism
  • KbinCafe
  • TheResearchGuardian
  • SuperSentai
  • feritale
  • KamenRider
  • All magazines
    •