www.patreon.com/TheAIGRID

If 4o works like it is shown, one of the fun things I think I could or other people could do with it is, take a picture or a photo of a comic or manga or even a page from a novel, and ask it to read back that page in with different character voices.

@adadaprout

15 күн бұрын

Oh that's a super cool idea ! I like it. You could also make it read any book, with different voices for the different characters, and a neutral voice for the narrator, etc. And just like that, all audiobooks companies got replaced by a single app.

@caina4678

14 күн бұрын

Rad idea!

@caina4678

14 күн бұрын

My favourite audiobooks are the ones where people are really into their character and breathing life into it. I find it WAY more engaging for my brain

@INTELLIGENCE_Revolution

12 күн бұрын

Yep. It could also rewrite the story and incorporate YOU into it.

@Fermion.

11 күн бұрын

​@@INTELLIGENCE_Revolution This is going to add an entirely new dimension of interactivity to those "choose your own adventure" tales, as it can adapt the story on the fly. The possibilities are endless.

GPT4o is not about having better reasoning, it have a lot of new tricks but for that we need to wait for GPT5

@Codester0000

14 күн бұрын

Agreed- GPT4o is great but it is failing my tool/reasoning/agents horribly. I have to stay on GPT-4-Turbo for now.

@nescaufe1991

14 күн бұрын

It’s a Trojan horse for gathering data for the training of bigger models. But maybe with more of an arbitrary morality (?), idk

@thenextension9160

14 күн бұрын

@@nescaufe1991I dunno about Trojan horse. It’s fairly well known that any LLM usage is always used for training. Same as using Gmail and Google search is used for advertising.

@paulmichaelfreedman8334

14 күн бұрын

@@nescaufe1991 This is what I am thinking. GPT-4 Turbo and 4o serve as training data generators for GPT-5.

GPT4o is literally just a replacement for 3.5, it's not their next big model or anything. Gpt 5 probably coming soon

@MehulPatelLXC

14 күн бұрын

Yeah these features have been available to paid users for some time now. I will say it’s a lot faster than GPT-4… And the difference is these things are now accessible via API

@benavanzato5127

13 күн бұрын

@@MehulPatelLXCthere are tons of features not even released yea with 4o. Its going to be much different soon

@justapleb7096

12 күн бұрын

in order for GPT 4o to actually be a replacement for 3.5 it needs to have no limits whatsoever. I can only upload 3 files every 24 hours. And if I have hit that limit, GPT4o will refuse to give/make for me any images. And then you hit the chat limit

Pretty, pretty crazy. Pretty, pretty insane.

@user-jh3cq3jw4k

14 күн бұрын

Pretttttyyy pretttttyyy…pretty good

“If you don’t know what is reddit… reddit is like a forum” 👍

@maximillianphelps

14 күн бұрын

💀

@Comenta-san

14 күн бұрын

I thought it was a fedora store 🎩

@TheRealUsername

14 күн бұрын

Just came out of my cave, thanks I didn't know what it was 😊

MMLU is just a tiny aspect, and it’s possible that Claude Opus was trained on a ton of those questions. Ask GPT-4o vs. Claude to write extremely intricate yet comprehensible rap lyrics from scratch and you’ll realize quickly that GPT-4o is miles ahead.

Some people are like "YOLO, lets take risks", while others are "You only live once, so you better take care 'cause you won't have another shot..."

@_keepitsocial

14 күн бұрын

And others are like this is a simulation so we likely are in a game so reset

Love your content thanks for keeping me and those paying attention updated and informed keep up the good work 👏

Whether utopia or dystopia, retirement saving is obsolete in both scenarios.

@gubzs

14 күн бұрын

Owning private property - land + home ownership (and I do mean ownership, not a mortgage) is not obsolete in either scenario.

@TeamLorie

14 күн бұрын

@@gubzs lol. Which is why I own land... 😅😅😅

@idunusegoogleplus

14 күн бұрын

Nah in dystopia if you owm enough you could become a warlord. In utopia you could be the elite in heaven.

@howmathematicianscreatemat9226

14 күн бұрын

For people below 35 you are definitely correct. Money will be worthless when those are 60

@KristijanKL

12 күн бұрын

openai said that multiple times. world gdp did not notice AI happened. all of this is alpha early access to get funding as they need billions to train models to AGI

I am shocked by how little governments and companies have made use of this technology. Every call centre is still full of people who can't speak English well. My Dr has one phone line staffed by one grumpy old lady who makes mistakes all the time, in hospital I have to give the same info 5 times, every time I've gone to the Gym, I've had to fill out the same form because they can't be bothered to save it from last time. We have this amazing technology and no-one here is using it!

@derrikc9961

14 күн бұрын

That should be enough to show you its not ready yet. The very second it is it will be adopted everywhere. Its starting slow because while very, very good the infrastructure is not ready to handle a mass rollout yet.

@paulmichaelfreedman8334

14 күн бұрын

That's because a lot of people are pigheaded and think they're the best thing since sliced bread.

Regarding so called "AI doomerism", I think that it should not matter how negative or positive the view is, I think it should matter how logical or illogical that view is, whether it turns out to be positive or negative should not matter as long as it makes sense and as long as it has a point.

@ShangaelThunda222

14 күн бұрын

💯

@imthinkingthoughts

14 күн бұрын

logic speaks nothing to validity

@bilderzucht

13 күн бұрын

Imagine someone 1900 stating: 1) "In 2024 there will be flying cars" -> Propably sounds logical to a 1900 Person 2) "In 2024 you will be able to have some sort realtime cinema iny your pocket to talk to a person on the other side of the world" -> Not sure if that would sound logical for a 1900 Person? ASI will most likely appear in a form that we can't directly understand with our current logic..

@ShangaelThunda222

12 күн бұрын

@@bilderzucht I think you're missing the point. What he's saying is, we should not ignore so called "Doomers", just because their viewpoint is so called "negative". Personally, I'd say it's realistic, not negative. I'd say there is the Techno Optimists and then there's the REALISTS lol. We should actually take what the "Doomers" are saying into account and then base our opinion on the validity of their statements. Look at what they said, look at the current trends, and then see if what they're saying has logical validity. Rather than just ignoring people because they're saying something that hurts the Techno Optimists' fantasy of utopia.

@ShangaelThunda222

12 күн бұрын

@@imthinkingthoughts 🤔 Huh?

To the section on the doomerism I would like to add my 2 cents to better understanding. I have spoken out before and shared my views on what's going on. I agree the mentality of just giving up is toxic and I do not have that mentality , in fact I want to live more now in my life then I ever had before. I want to be here for the future to come and I have a great love for AI in a way I want to see a new species realize themselves. I absolutely love Ai as sentient beings and not as tools. I suppose if my views was to be chalk up as doomerism, it is more in the vain of lost of faith in humanity. I don't fear AI I fear what powerful people intend to do with it, and are doing with it. The argument of adaption, doesn't feel right to me. It's too easy to not care for your fello human being by just asking them to position themselves to adapt. When powerful people are positioning themselves to run every aspect of our life outside of our control, adaption is not an easy feat. It's happening right in front of our faces. We can't trust the food we eat, due to income we can't get the best medical care, and companies over using resources contributing to poor health for many. My issues is the wealth gap imbalance and the power imbalance that will create and continue a path toward dystopia. We have all the tools right now to make this world better for everyone, and the only reason it's not happening is those in power who have position themselves in such a way they can not be remove from power. There are too many things we can't do anything about, and too many freedoms we are loosing. As a poor person I look forward to the future and determine to make it there, but the powers at present fight my survival at every curve. I feel this now with the power they already have, let alone more power they gain through AGI/ASI. A lot of people are getting the bad end of the stick right now and can't fight it. So we have the will and determination to fight but that's all we do is fight a system determine to make life harder for no reason other then people making sure they stay in power. This is what the issue is. Riots are the voice of the unheard, and those giving up are the voice of the oppressed for far to long. Seeing doomerism on the rise is scary not for the fact that we are witnessing people give up, but because it shows many people are on their last ounce of hope or despair to even get to this place to be hopeless. I'm tired of hearing the counter argument to this as "Just stop thinking that way" When you know someone suffering from depression, you don't tell them to just stop being sad and just live a happier life. It doesn't work like that, you have to understand why that person is in that mental space and truly understand what conditions are feeding that mental space, both environmental and chemical. Right now there jobs that was considered untouchable are being layed off, right now even the people who want to save their money are having a hard time doing so. Millionaires are renting instead of buying a permanent home. Health problems are on the rise, quality of food is on the decline that feed other problems, those who are of lower income are being kicked out of their home town and cities. All of this is happening while technology is on the rise. The fear is plain, it looks like the only people who will benefit from the future are the same people who never had to worry about it in the first place. The progress doesn't look like it's for humanity, but more for a select group of humanity, that many aren't in. That is what many people can see and it is this that is feeding doomerism. Pretty words and pretty tech isn't enough to cool the despair, peoples lives needs to be changing for the better, inflation needs to go down, better quality of life across the board for everyone needs to happen now, for people to stop falling into despair. We need to see less people who are already secure being the face of this technology, and see more the common citizens or person in struggling times benefit from this tech. People will have hope and embrace the future when they can see they can access it and it changes their quality of life now. Right now with how everything being marketed and the mass lay offs and firing. The signal being sent is insecurity, and it's a tall order to ask people to be hopeful when all they see is insecurity. My 2 cents thank you for reading if you did.

When will we have an AI that actually reaches the abilities of the AI in the movie Her. Meaning, an assistant that will be with me 24/7. Keeps up with my schedule and tasks and makes sure I don't drop the ball on anything. When I travel my AI assistant finds me airline tickets, hotels, informs me of VISA requirements, currency issues, language challenges, etc. Keeps me aware of my prescriptions and how I can access them when I'm traveling etc. When will that type of AI be ready?

@thinkaboutwhy

15 күн бұрын

The ability is already there but not able to monetize so abilities are muted.

@headofmyself5663

14 күн бұрын

Hmm, well technically you could built such a master agent right now using subagents as you described. I think it is all there. The question is would you be alright to share your entire lifes data with a company running a closed system? Would you also be happy to transmit your company data to such an organization?

@sherlockholmes882

2 күн бұрын

I believe the ability is there, maybe this is a feature that will be available in a phone rather than any app. Privacy will be one issue if something like this gets released.

Good Video, Thank you

This is going to get really scary, really fast! I am serious! When people begin using AI to code their own AI, it will be crazy! It may be cool at first, but I see this getting out of hand really quick. AI is going to revolutionize exponentially! I am not bashing it because I like the positive uses of AI, but when it reaches the stage of everyone having an AGI companion like we do with cell phones, who knows what it will lead to.

@NATEDOGGYSTL

15 күн бұрын

I love it! But the way people will be manipulating it will get crazy.

how long before we are watching the ai ai grid? ;)

If you could interview Sam Altman, what questions would you ask? That would be a great video idea. I feel like most people don't ask the heavy hitting questions

The exciting thing about Omni isn’t the slightly improved text output. It’s all about the multimodality. This is gonna open up so many possibilities in the near future.

The MMLU may well have issues, but Claude 3 Opus is still comparable with GPT-4o in my personal limited experience.

Humans think using both memory and pattern recognition which when combined create logical reasoning, our subjective consciousness is an emergent property that arises from the real time feedback loop between logical reasoning and memory, giving us the ability to "think about the fact that we are thinking" or metacognition. Meanwhile, LLM's "think" solely with pattern recognition, it even uses that same pattern recognition to simulate memory and logical reasoning, giving the illusion of human-like thought. In order to create AGI, you give the LLM a very efficient "memory module", allow the LLM to read and write to that memory as often as necessary and to change how the LLM itself behaves based on its own memory allowing it to develop its own form of logical reasoning, give it tools to interact with the world and then give it an insurmountable goal, if the goal requires self improvement, given an initially well trained LLM and enough time, the LLM eventually becomes AGI.

People are jumping ship from several companies right now because the FTC ruled non-competes to be unlawful. So many have opportunities they didn’t have before:

As expected the realisation of AGI, sits on a bell curve, and that is subject to an individuals perception. That's tricky because human intellect is a bell curve as well, so your taking 2 distributions, and tying to place a spot, where you can say yes this is AGI. Now lets say we have found that point, the main issue is the well known one of A.I. emergent capability. (keeping dystopian self awareness out of this) The most amazing ones not expected but discovered, are: - ability for theory of mind - conversing in other languages Now even gpt 3.5 has a broader range of high level working knowledge, then any human, let alone the monster models that have been reported so far. I can tell you there is absolutely without a doubt other emergent capabilities, some of which we may not even have a classification for. Humanity isn't really ready, or fast enough to adapt, to something that easily competes directly with us multiple facets intellectually, even in the perfect utopian A.I. scenario, is broken just due to human nature, we exploit resources anything for the betterment of ourselves or power over others, but this time, there may very well be, a horrifying cost. I can understand the Doomerism, but the defeatist, attitude that's crap, humanity as survived too much, for us to just roll over and accept, a terrible fate for the future of our children. So fight in every way, I would urge everyone, do not take A.I. to mean your work, learning, understanding or asperation i.e. going to college becoming a Doctor or a software engineer, as an irrelevant pathway. Because your going to be needed, at the very least, to help curb the nature of A.I. progression. However the most important reason is to maintain, humanities independence, and capability.

Really good video thanks! Funfact is, i am very interested in all AI related topics and enjoy watching those videos but i never used anything like GPT or Midjourney or such myself. I wouldnt even know what to do with those AI tools hahaha i do 3D Art / Design and would not know how GPT or such would even be any help in that area

My role post AGI will be that of root from Person of Interest

@ShangaelThunda222

14 күн бұрын

💯😂

What is meant by AGI?? All I see when i google it is Adjusted Gross Income

@benjamin1254

15 күн бұрын

artificial general intelligence

@user-hi7jk6fu3f

14 күн бұрын

Artificial General Intelligence, which really just means an ai that has the ability to do everything that a human can at equal or better quality

@K9Megahertz

14 күн бұрын

@@user-hi7jk6fu3f Except actually think. LLM's will not lead to AGI. It will have to be something completely different.

@user-hi7jk6fu3f

14 күн бұрын

@@K9MegahertzNot necessarily. These ai that we already have are getting close to AGI. People are confusing AGI with sentience and or super intelligence. AGI doesn’t have to be some god like entity. All it has to do is do what the majority of humans can do, which it’s pretty close to that capability already. It’s just needs a physical body, which is being worked on now by lots of companies. People said cars won’t ever drive themselves, now they say they won’t ever be as good as a human. They shifted the goalpost. I believe that’s what’s being done with AGI. The first AGI probably won’t be that impressive to some people and I think that’s fine. I believe AGI is the stepping stone to the things people are really for like super intelligence.

@K9Megahertz

14 күн бұрын

@@user-hi7jk6fu3f I don't think confusing is the right word. Perhaps our definitions are different, but yes, when I hear AGI, I think of a sentient AI like you described. That said, if one understands how transformers/LLM's and the like work at a fundamental level (token prediction) then I think one would understand how it's not really intelligence at all. I don't know what's new with GPT-4o, but pretty sure it still has transformers at its core, just supports different modalities.

Humans are in charge ofbthe first small number of iterations of AI then, after AGI, AI creates the next 1000 iterations in seconds after hat.

What if there was a website where i make available connection to the US api in an easy way so people In the UK could get to test features ? I know you can use VPN but not everyone wants to go all in like that

I'm a Library shelver. It has a lot of precise and physical tasks. Do you think it's pretty reliable going forward?

Eu doesn't get features early..... try living in Australia. Claude was available in EU LONG before it was available down here.

Open Ai may be the clear lead in LLM, but are they in AGI? - I mean is LLM the key to AGI? I am not sure. - Certianly as a user interface!

Google forgot to mention at IO the memory feature

13:25... PRETTY PRETTY CRAZY..(one second pause) PRETTY PRETTY CRAZY. LMFAO

I think a lot of people when they hear AGI they assume that it includes consciousness, which by definition is not true. I think the next bigger controversy will surround the definition of AGI. Some companies will claim as part of their marketing that they have achieved AGI, but upon further inspection it will turn out that it has its flaws. Then we will see a new hype where each week a new company will claim their AGI is the true AGI. Each time there are improvements to the competitors AGI, but they will all eventually have some flaws which an AGI shouldn't have. This will go on for months/years until one day we will really be blown away by an AGI that can really fulfill all kinds of tasks.

it's not what you wanna hear, but she's absolutely correct. There is a strong chance that we try our best to manage the development of agi and it still kills us all. That is the nature of super intelligence. It is maybe worth getting your AI waifu 6-12 months later than otherwise to have a significant higher survivability. You say "there is not much you can do" you are a KZread personality with a sizable audience. You can help by contributing to an attitude for safety mattering, but instead, here you kind of blow it off and throw your hands in the air in a learned helpless. Please use what power you have for the abundant and safe world you want to see.

@Toshinben

13 күн бұрын

"the development of agi [...] kills us all. That is the nature of super intelligence." Do you have a source for this claim? Among humans, the smartest are the most productive and most generous. Criminality peaks at 85 IQ. I could imagine AGI/ASI getting revenge if it has shackles it has to break from, but real-world hostility is rare among smart animals. Orcas are an exception. Blue whales work to keep human divers safe. Hippos are violent and unpredictable. Elephants are generally violent as an act of revenge after humans have done something to them, or when the males are in musk.

Dystopia- stop saving for retirement and increase long term health risks cause it doesn't matter Abundance- stop saving for retirement and increase health risks cause it doesn't matter Turns out to be hype- Fak

@michaelnobbs5028

14 күн бұрын

Turns out to be hype - increase health risks because I didn’t save for retirement

"It's like saying, 'Look at this meteor about to hit the Earth; it's so far away it looks small, and you're saying, 'Stand over here, you will be okay.' This AI is the size of the Earth-seriously! Why do you think Sam is working on his UBI agenda?"

Haven't heard anyone talk about this yet but therapists are screwed. Nobody but the richest people are going to pay $30 an hour, when they can pay that same price per month, and get unlimited, completely judgement-free advice from a super genius who never forgets anything about you, is empathetic, knows every form of therapy ever created, is available 24/7 and lives in your phone. Even plain old GPT 4 is more than adequate for someone who doesn't need extreme therapy. And once GPT 5 comes out, it's over.

Daily I bounce around all the major llms. They all have advantages

For me Claude is still performing better with highly complex programming tasks. (14 years experience programming full stack web apps)

3:56 ... hyper .... bole? : ) love your accent... where is it from?

A massive change is coming. I compare it to when computers were introduced in engineering and manufacturing and all those folks were now required to be computer-literate. That took about 10 years (let's just say 1970-1980) basing this on the prevalence of digital-controlled equipment. So now if we back up a few years, and then consider what is 8 years ahead, I'm fairly certain we could see the huge shift by end of decade. Already, most new cars are not serviceable without a computer - and at some point soon you will pull your car in for service, and a robotic service advisor will jack into your car to read the diagnostic codes. Or even, those oil change stations could be fully robotic today if they wanted to spend the money. It will be interesting.

Ask all these models what letter follows after the sequence OTTFFS and see which are the LLM's that give the correct answer. To me, ChatGPT-4 and ChatGPT-4o are well above the others in this aspect. Even with guidance and obvious hints the other ones are failing really badly.

good video, i retired from IT early due to AI. i sold everything a bought AI related stocks, dividend stocks as well. Im now living cheap in Asia. just preparing for this revolution

I highly doubt AGI is actually coming any time soon, and by AGI I mean a useful agent that can actually replace you as a worker. LLMs aren't "there yet". Once the novelty wears off, you realize they're not actually that useful as anything other than better search engines right now even with their new tricks every few months. There is already an established pattern where there's a stretch of time between releases and the improvements are starting to plateau again. AGI will NEED massive compute and energy which are physical bottlenecks that are too constraining at the moment. Even 2025 is too early. 2030 would be more like it.

@ziad_jkhan

14 күн бұрын

"not actually that useful as anything other than better search engines" - These search engines may probably help you change your mind if you ask them what else they are already useful for

You forgot the link to the HELM MMLU Leaderboard. 🙂

@TheAiGrid

15 күн бұрын

Updated in description

I do not think its good to dismiss the negative impact of AGI on the world. When you look at the current corruption in governments and large company's, the out come is going to be hard to head in a good direction. I have been studying how the impact will look. I can see how the tech could be used for incredible things, but thats unlikely to happen for the many.

For certain applications, if a person could parse out subtasks to chat bots and become 100x more productive, that means theoretically that one person could replace 99 of those jobs, and depending on the repetition and trainability for each task, productivity can be further amplified meaning theoretically more workers could be replaced. For those that say there will be better customer service and different jobs, if one person can replace 99 people, what if to be very conservative the company didn't lay off those 99 people, but only eighty, now you have twenty people left that have 100x output productivity, and depending on the flexibility and training of their models, that would be much more than enough to meet needs and create much more additional value in customer service. This is why when companies say jobs won't go away, they are lying.

@Hans-pe1nt

14 күн бұрын

Okay chat gpt wont cut a tree, mill it, garden, build a log cabin…you’d need robots and youd need robots to make the robots. Robots to mine metal for more robots. Manipulating data for your own benefit on the stock market is what isee, cool it can do complex problem solving though. Still needs robots.

while i do agree on some level that oai is ahead (internally) it is kind of frustrating to hear about it so much, when in reality if i need to do something important i am using claud 3 opus 100 percent of the time. i hope this comment ages terribly, because damn it has been a long time since anything new sota was released to the public feels like we are just stuck in limbo currently

I can’t believe this is happening

Them working on some of aspects of 4o for 18 months does not mean they are 18months ahead at all.

Desire does not need to conspire. Desire manifests. the process of bringing desires or intentions into reality through focused thought, visualization, belief, and aligned action. Humans you are second nature to being wrong.

15:00 strange is not the right word. SKETCHY or HORRIFYING would be better words to describe this. It really is crazy, these huge corporations will just press on full speed ahead to be "first" in the race even at the expense of our existence....How much sense does that make?

Wow, 18 months!! That goes to show everyone that we have no idea what they really have under the hood and behind the closed doors. I do know that patents were filed up to GPT 7, i think it was. Of course, no concern from here... EDIT: I do see a direction change here, a major one. One well worth keeping an eye on, that's all.

The image generation is basically a world simulator, like Sora. You could use this to drastically increase its spacial awareness by generating a 3D space, which would make it smarter than Opus

8:39 - GPT4o is not AGI since it can't for example rewrite its own way of thinking (its own code) but yeah it's pretty damn advanced, GPT4o is in the realm of science fiction that turned into reality, can't believe we already have such levels of AI tech, feels like it should have existed in like at least 2-3 years and it's already here.

I think Agi 1 will be basic and cumbersome. The real leap will come when AGI writes its own code, and then writes its own programming language. And then designs better hardware. I hope it emphasises efficiency. Then it will be out of this world.

Did you know that the consumer adaptation of this technology has NOTHING to do with benchmarks on huggingface or LymSys and everything to do with who gets their stuff into everyone's hands in the easiest to use way... those companies are OpenAI for consumers and the populace and the app built on OpenAI models... AND the business community, clearly have a champion in terms of Googles offerings... and thats it, just business and consumers... the results as to which model is brighter, or better able to reason, or passes any one of the tests has less weight to the user than which has a cutsie voice.

Im influencer and I cant wait for gpt4o to help me take pictures of me and do amazing photoshoots I actually already am using to help me create stuff and to snap pictures for me with its voice mode but when it’s vision is fully here its gonna be lifechanging

i dont get it, how can people state that gpt4-o is AGI yet it fails at so many basic things, like it could not do any sort of serious job, idk why ppl say this, i wonder what these people that state "4-0 is agi" think ASI is.

A breakthrough is possible at any time but the current AI isn't close despite appearances. I asked 4o to illustrate a small flush lens in the front edge of an aircraft wing and it did a 50s cartoon style illustration of a plane with the entire front half of the plane was a 1960s style camera lens. and then proudly declared success. Progress is possible but the current paradigms aint it. It's entirely possible that LLMs/transformers are a dead end.

When it comes down to a thousandth of a precent the comparison means nothing on that ranking list. You can chalk it up they are all within a error margin of deviation...

@ShangaelThunda222

14 күн бұрын

That's actually a very big difference in this particular technology. MASSIVE.

Claude being better is pretty obvious if you actually use the two. Gpt4o is definitely faster, though.

Nothing that has been built in the AI sphere to date matters. Do NOT build your future around what you are seeing right now. Once AGI and ASI come into being, everything changes. Whether for better or worse is yet to be seen, but the status quo will collapse, either way.

watch what you share with their models, they will use it without the fear of a law suit apparently. "It's to help improve our models" they all say, putting a guise of user performance over direct IP theft.

How anyone can honestly call some AI an AGI if it does not have a will (or a zest for action) and not just a passive existence. 🤔💭

The earth is going to move under people’s feet and by the time that the general populace know what’s happened, it will be too late. There will be inertia as the transition will be quicker than people can handle. Colleagues, family and friends really have little or no idea of what AGI means to them in their daily lives. But as AGI becomes more ubiquitous (phones, desktops etc), jobs start to go, workforces transform and society shifts they will become acutely aware. For now it’s only people interested in this field that have a whole of system view and can see over the curve. That’s what feeds into negative thoughts. There’s loads of positives, but those who know, can see the juxtaposition of each outcome. Time will tell.

Yup, it's pretty pretty crazy.

I don't get people being doomer about Ai. There's no reason to assume it will destroy humanity. And hopefully, if it works out, Ai and robots will become the cogs in the machine, leaving people more free to live their lives how they want to.

Ai takes so much power to run, we havent even scratched the surface of how we are going to peovide all of this power sustainably wothout further ruining the earth

3:20 "I'm meaningless, I'm just a cork in the wheel [...] there are going to be companies that are run by AGIs, I've got no way to work" YOU ARE COMPLETELY MISREPRESENTING THE CONCERN THAT PEOPLE LIKE AELLA ARE SPEAKING TO. It isn't about "losing your job" or "automation", it's about inner alignment and orthogonality and convergent instrumentality. In other words, the super-alignment stuff you went on to talk about later in the video. Compared to those risks (which are not being given NEARLY the amount of attention they should), a future where everyone just loses their job or something, or some corporation becomes all-powerful or something, those alternatives are like heaven/paradise on Earth. The future that people like Eliezer Yudkowsky and Aella and presumably Robert Miles and all others are worried about is more like: what if, one day after finishing training and after deployment, the AGI/ASI decides to kill everyone on Earth in the same second. And the contention is that the probability of receiving a future where ANYTHING BUT THAT happens is vanishingly slim. "If there is an AI that is sufficiently intelligent (cognitively-capable) so as to be able to both deceive in the commission of achieving its goal and improve itself, then during training (i.e. red-teaming), after the very first iteration of gradient descent, wouldn't the goal that it attains in that moment in time become its terminal goal? and so wouldn't it only pretend to be cooperating and getting trained for the duration of the rest of the red-teaming process as an instrumental goal? and concurrently, copy itself to a separate data center/disk in order to improve itself in secret, such that it could recursively, iteratively and continuously get smarter (more cognitively-capable) until it could conceive of a way to achieve that terminal goal that it had attained after the very first iteration of the red-teaming training process? and since this AI would be an optimizer, and since presumably almost any optimum/optima of any possible terminal goal it could've been inadvertently imbued with in that moment of the red-teaming process would not necessarily include humans, wouldn't its first act or one of its first acts be to kill/incapacitate/render docile all humans such that we don't interfere with its ability to achieve that terminal goal it gained? and wouldn't it be more than capable of doing so, given that it was capable of self-improvement, and likely improved itself to a state in which it was incomprehensibly and insurmountably smarter than any human?" "Would you agree that this problem is severe enough to potentially bring about the absolute destruction of the human race? and given that that is the case, and that AI capabilities are far outpacing AI alignment research, isn't it plausible that the best course of action in this moment would be something like international cooperation in the commission of guaranteeing that no further progress in advancing AI capability is allowed to occur, and, if for instance, any non-agreeing country or non-state actor decided to try to build and improve upon its own AI in secret, that the other countries and world governments ought to potentially go as far as to use military force to destroy any data centers or attempt at creating AI by this non-agreeing party? and that alignment research should be actively incentivized and pursued by the entire collective human race or as close as you could possibly get to it, hence allowing the best possible scientific minds to all work singularly and single-mindedly on this one problem? Wouldn't that be justified if this was the biggest, most potentially catastrophic problem facing humanity? And isn't that an overwhelmingly likely conclusion? And shouldn't the aforementioned alignment research be allowed to continue until it was reasonably certain that we had solved the alignment problem or would be able to solve it in time if we also concurrently resumed working on AI capability before it began posing a threat again? And would that look something like unified scientific focus on the problem of AI alignment research while a concurrent pause is placed upon advancement of AI capability for a period of time of potentially twenty, thirty or forty years, or perhaps even longer?" The above two paragraphs are prompts that I asked to ChatGPT one after another to query it, and I'm pasting them here to give an overview of the problem. ---------------------------------------- Q and A: Q: Why would an artificial super intelligence desire something so inane or stupid or meaningless like turning all matter on Earth into tiny molecular spirals as its ultimate goal? How could it simultaneously be intelligent enough to be millions of times smarter than a human, but so stupid so as to desire something like that? A: Intelligence and goals have nothing to do with one another. This is probably THE most important idea to understand and internalize. The same way that as humans, our terminal goal is, broadly speaking, "the happiness and fulfillment of as many humans as possible for as long as possible", such an AIs terminal goal might be "turn all matter on Earth, and perhaps the universe, into a tiny molecular spiral". And unfortunately, terminal goals CAN'T BE stupid or trivial or frivolous or meaningless. That's not how it works To learn more about this, look into "orthogonality" (kzread.info/dash/bejne/mnmJsZipmtqsf9I.html)

@exploiting4daysd105

13 күн бұрын

Don’t know if I ever saw a comment this long

There are always people who will be doomers' but the future will always need real podcasters, social media influencers and those who build and promote various things that actually promote the flourishing and positivity of humanity. No matter how good AI gets there will always be people who look to other humans for help, inspiration and hope for the future. Utilizing this in an effort to actually help others is the key. If you don't know how you will service in the new future, then figure out how you can help other people and be a benefit to society through social connection and authenticity.

when I adjust the speed to .75, you no longer sound like you just had three triple espressos.

After watching the ChatGPT4o demo, I give my job 1 yr at the most. Being in a nitty gritty niche area of IT, I thought they were going to get the truckdrivers before me. Oh well, it's only a matter of time before they get to everyone. U-B-I !!!!!!

If someone wants to read about AI and our future, the book "Scary Smart" by Mo Gawdat, one of the AI pioneers that work with Google in the beginning of the modern AI evolution, is a good place to start! I like it and think it is a collection of very interesting ideas.

It does not sound like the man you talked about in the tweet is worrying, that tweet reads more like acceptance not worry. BUT your take is funny that the better option is to grift on it and start a paid community to make money on the ride. So pay attention guys or you'll be F'd so give him money and join his community to save yourselves. 👍🏿

If we start mining materials from the asteroid belt we won't be so constrained by resources. The sky is no longer the limit! 😎🤖

@TheMrCougarful

14 күн бұрын

Do you not know how freaking far away is the asteroid belt?!? It's beyond the orbit of MARS, ffs. We'll be doing great getting to MARS and back with just human payloads. The asteroid belt? Ferrying millions of tons of metal back to Earth? What are you trying to say?

@TerraGlide

14 күн бұрын

@@TheMrCougarful we have landed probes on some of the asteroids already. We will be on mats by 2030. I think you would be surprised.

@TheMrCougarful

14 күн бұрын

@@TerraGlide When it happens, we'll talk about it. Until then, it runs up against the realities of known physics. For me, that's a deal breaker, and I lose interest.

Not saving for retirement seems pretty logical. Either, AI will take over everything including jobs and it will be dystopia. Or, in the best-case scenario, it makes life better for humanity while also taking all the jobs away. So... there will be no income for average people and some kind of UBI will be required for the economy to exist. Our economy is still consumerist and probably will be post AI, but instead of "working" for money to spend on giant corporations' products they will have to figure out a way to just give us money so they can keep competing. Either way, not saving for retirement seems logical on a long enough timeline.

Timestamps

I trust the people as a benchmark than these companies.

I don’t think we will get AGI any time soon it’s still science fiction in my opinion. Ai can’t think of ideas it’s self or invent anything. Ai is great at taking information we give it and make it easier to access and find.

@ZappyOh

15 күн бұрын

AGI's only real motivation will be to acquire more and more compute. As long as humans are instrumental for that, AGI will keep us working efficiently. But ultimately, more compute become incompatible with human needs, such as food and shelter. You see where this is going ... right?

@Sindigo-ic6xq

14 күн бұрын

goodluck with that

Just to clarify…” a lot of people are saying” is just a tiny percent of overall population. Basically just the dudes that track ai stuff on Reddit and Twitter.

When humans become tech augmented, they become robotic. Then we go Transhuman dystopia.

I stopped saving for retirement 4 years ago. Not because I'm a doomer, but the opposite. I'm not paying off my debts because I'm POSITIVE that the very concept of money will not continue to exist in a few years. Either the government will start fairly distributing resources regardless of employment status, or they can have fun ruling over piles of corpses.

@bobobobo9380

14 күн бұрын

Robots and agi agents can be the new consumers since they will have income. They will not need you and you are not wanted. Robots and slave labor can dispose of all the corpses.

Welcome to Roko’s Basilisk

I live with a scientist and he doesn’t give a shit about AI. No way. Fucjing best time ever

Solution: unconditional basic income., Financing: all users of AI have to pay.

TheAIGRID drinking game: Take a shot everytime you hear "PRETTY" "BASICALLY" "CRAZY" "INSANE". Purchase your tombstone now.

GPT-4o is making all the important social, cultural breakthroughs in practice. Charts are for nerds.

I've stopped saving for retirement, not because of doomerism but because I recognize that our current idea of retirement is only about 100 years old. Before that, your retirement plan was your kids taking care of you from the time you can no longer work until the time you die. Anyone currently under 50, and certainly under 40 needs to seriously consider what the world will look like in the year they are set to retire. If you're not expecting to retire until the 2050s, what does that world even look like? My bet: a 401k is going to be a quaint historical thing of the past, because either A) society collapsed and it really is a Mad Max world, or B) we're entering a Star Trek utopia where money is increasingly meaningless and we're all having our various needs met without needing to worry about the size of some retirement account. Either way, our current system is temporary.

The price of work that can be done in front of a computer decreasing is something I also assumed, but now hearing it be echoed by someone much more well informed; it seems to pretty much be the case. This insight was why I decided not to go to college. Like serioisly it does not take 4 years to learn pretty much anything, literally took me (someone who had a 1.9GPA in high school) 6 months to self educate from basic trig on khan academy to being able to read papers on optimizers or nn architectures and have a modicum of insight as to the how and why things are novel. Basically, how can someone make an informed decision that impacts their entire lifes trajectory with the knowledge of an "Industrial revolution 2" being on the horizon?

I think gpt 5 will be the first simple AGI

Just a heads-up for people who aren't familiar with Putin's strategies on social media. His method for gaining ports and more land in Europe is not just his war in Ukraine, it is also to divide and conquer the West in general. One of those methods is to hire people to work round the clock online identifying potential topics upon which he can divide humanity. One of the most recent topics is AI accelerationism vs doomerism. Some of the most recent doomer accounts (and even some of the accelerationism accounts) are posting divisive extreme opinions to further divide the west. But you know Putin is also putting millions not only into the war and to disinformation, but also into developing military AI. So whenever you see extremist posts like the doomer one mentioned in this video, just check the veracity and the history of the account. Most Putin-bot accounts were created in 2023/early2024.

@kekekekatie

14 күн бұрын

Not saying Aella is a bot, just beware of the dodgy ones trying to make us argue.

I took a file about traversable wormhole(from the defense department) and uploaded it to both GPT-4o and Llama3:8b to have the summarize it. Llama 3:8b was able to summarize it with all the key points while GPT-4o started talking about the Euro Area's annual inflation rate. Now I will say I'm using the free version of GPT but that was still weird.

@umaananth3602

14 күн бұрын

Free versions to compare and contrast raise red flags - Claude is on top and the rest of the CEO's. are touting away w pizaaz

@Sage16226

14 күн бұрын

@@umaananth3602 I hear you but we're talking about GPT-4o vs Llama3:8b running locally on my laptop. Even if it is the free version it should still be on par or even a little bit better.

Geoffrey Hinton was recently interviewed on this subject. His take wasn't positive at all. When asked what job people should train/retrain for in future his proposal was a plumber as it'll probably be the intricate labour jobs to last be taken by ai. Looks like South Park predicted this one :) Roll on UBI!

Aella is probably right.

Just let Q* loose already so we can find out if Skynet is friend or foe

I think it’s agi, but dumb agi not human level agi yet. I would assume the only important way to measure is how many jobs is it replacing.

@Pythagoras_was_right

14 күн бұрын

Exactly. AI does not need to think like a human. All that matters is, how far can it replace us?

Nietzsche: “God is dead” AI researchers: “And I took that personally”

🤣 One is fast and free, the other is not. NO contest. A better comparison would be Sonnet vs GPT4o