# Sam Altman REVEALS AGI DATE In NEW PREDICTION (AGI DATE!)

## Метаданные

- **Канал:** TheAIGRID
- **YouTube:** https://www.youtube.com/watch?v=WVs6PXKiKmg
- **Дата:** 11.03.2024
- **Длительность:** 27:15
- **Просмотры:** 30,487

## Описание

✉️ Join My Weekly Newsletter - https://mailchi.mp/6cff54ad7e2e/theaigrid
🐤 Follow Me on Twitter https://twitter.com/TheAiGrid
🌐 Checkout My website - https://theaigrid.com/

Links From Todays Video:
https://twitter.com/sama/status/1520798948562141184?s=21&t=Li4AGEyohNxr5hmw4LPAtw
https://www.marketingaiinstitute.com/blog/sam-altman-ai-agi-marketing

Welcome to my channel where i bring you the latest breakthroughs in AI. From deep learning to robotics, i cover it all. My videos offer valuable insights and perspectives that will expand your knowledge and understanding of this rapidly evolving field. Be sure to subscribe and stay updated on my latest videos.

Was there anything i missed?

(For Business Enquiries)  contact@theaigrid.com

#LLM #Largelanguagemodel #chatgpt
#AI
#ArtificialIntelligence
#MachineLearning
#DeepLearning
#NeuralNetworks
#Robotics
#DataScience

## Содержание

### [0:00](https://www.youtube.com/watch?v=WVs6PXKiKmg) Segment 1 (00:00 - 05:00)

so there was a recent statement by Sam Alman that actually revealed his prediction for an AGI deadline slate SL arrival or whatever you want to call it and I think this video is rather important because the timelines in which AGI is going to arrive is one of the most hotly predicted things that has been predicted in the realm of AI so let's dive into his statement what it means for us and how we can kind of gauge where AGI is going to be in the next couple of years in his prediction so essentially I'm not going to waste any time there was an article in which someone did an interview with Sam Alman for a book and in that article they talk about Sam alman's prediction now what's also good is that in that article there were actually a giant number of quotes that Sam Alman has actually said that we will be diving into in more detail now you can see here that Sam Alman has stated okay that AGI will arrive in 5 years give or take maybe slightly longer but no one knows exactly when or what it will mean for society so you can see that someone was asking him when it will be a reality and I think this is rather impressive because the prediction from someone like Sam Alman I think it's one of the most important ones because he is at the frontier of AI development and his company open AI you know the company openi which Sam Alman is the CEO of means that he of course I would argue knows the most about AGI technology in terms of where we are in terms of development because he's of course at the Forefront of that so his prediction stating 5 years give or take may be slightly longer but nobody knows exactly when or what it will mean for society is a very important quote to break down because it means that potentially we do have a date which can kind of give us some kind of sense when this AGI technology is going to be there now this is of course just his statement things are exponentially increasing and of course we do know that with AI development and with technology there are certain breakthroughs which can allow for companies to move faster than before for example with the Sora video paper um I'm pretty sure it was some breakthroughs at Google that actually led openi to developing that even faster because Google was actually quoted quite a lot in the Sora paper so um the point is that of course you know whilst his prediction might just be based on opening eyes knowledge and of course he's going to have you know pretty much every single piece of Industry Insight that there is the point is that some breakthroughs can come from other AI competing Labs that are going to be working on this technology also so this 5 years prediction it could actually get shortened if there are some breakthroughs because even with Geminis 1. 5 Pro I think there were some breakthroughs that Dem saus talked about that essentially said that look we've developed this technology and a lot of people think that we just scaled this up no we've actually made some new Innovations which we haven't talked about so there are some key Innovations which could be you know had in other labs which could definitely lead to some other breakthroughs as well and I think 5 years is a really good date and later on in the video I'm going to get to why I do believe that 5 years is a pretty important date based on some of the previous videos that you may have seen now essentially what's also cool as well is that um this entire article actually does have a decent bit of information and I think it's something that we should take a look at so one of the things he says was you can see that it says this is a lot of process to say the least it's a hard thing to wrap on Minds around says Roa but it's not as far-fetched as it might seem we don't need AGI to see masses disruption and transformation in the economy Workforce education and society and I think that is rather true and in this article the person who interviewed Sam Alman actually does quote you know colana because colana did recently hired an AI assistant that does the job of 700 um and I don't know they got rid of 700 employees they just said their new chatbot does the work of 700 employees so I'm guessing that it is going to make a lot of companies a lot more effective and of course we don't need to see massive disruption with only agis release we could of course just have it with Advanced AIS that aren't actually AGI but of course just narrow AI that's really good like for example large language models are really good at text really good at creative writing and those will you know impact certain areas of the economy that really will disrupt it so of course you know it's something that we're already seeing and what also this should you know keep in mind is that you know how Sam Alman has said that you know look this is 5 years give or take I think an important thing to predict is that even if you know AGI is that you know of course crazy system that can pretty much do anything a human can do it's important to know that one you know AGI isn't be all end all it's like there's spectrums of AGI like there's AGI that's you know half AGI you know I think some people would argue that gb4 or claud's recent model is Agi but the point is that timeline is going to be accelerating so much up until that point so when AGI arrives I'm pretty sure that we're likely to know beforehand not in the sense that we're going to know the reveal date but the point that like the kind of AI systems that we will have around will know that AGI is likely to be just around the corner because there likely won't be many breakthroughs left or maybe that will just be completely wrong but I think that AGI won't just catch us completely by surprise because of course as you already know we're already looking at things in their current state and we can see that things are moving pretty quickly now you might be thinking what is AI going to be able to do and in this article they actually did talk about you know what it's going to be able to do and essentially you can see

### [5:00](https://www.youtube.com/watch?v=WVs6PXKiKmg&t=300s) Segment 2 (05:00 - 10:00)

here that it says AI will handle 95% of what we agencies and creative use for today so it says Alman and open AI have stated many times that their mission is the development of AGI or artificial general intelligence typically we view AGI as that is broadly more capable than humans at a range of different tasks and in this book Alman takes it further describing AGI as when AI will be able to achieve novel scientific breakthroughs on its own and essentially it was just something where you know these people they are like marketers so they were asking you know how will AGI impact their um you know marketing firm but you know before we get into that stuff I really want to talk about this because essentially AGI being able to achieve novel scientific breakthroughs on its own is going to really accelerate the amount of research and development that we're going to be getting across many different Industries because if we can let's say we can just you know like I don't know how many times you can clone gbt 4 but like let's say you can clone it infinitely the only you know bottleneck is you know your compute essentially if we do have an AGI system that isn't that compute intensive we could basically have unlimited researchers we could scale up our researchers by a THX in any industry and then of course we could have them running 24/7 publishing research papers literally every single day every single minute um and the kinds of uh you know technological developments that we could have I mean it's kind of hard to predict what Society will look like if that breakthrough is achieved to where we get a system that can actually do that with things that do work effectively now there's going to be a lot of things that need to be done before that but um that's what Alman described it as and I think the way how they are building AGI I think that is going to be something that will be a key milestone for them you know for open AI for them to do because I think they understand how crazy that will be um and I I do wonder if they're going to develop like an alpha go type system which is just really you know focused on that kind of thing or it's just going to be a plain AGI system that can be used across everything and then just do scientific breakthroughs so it will be interesting to see how that does work and then of course this is where they talk about how you know this AI system could be used for marketers so he says you know how will it impact their world of marketing and he says you know what do you think AGI will mean for us and for consumer brand marketers trying to create ad campaigns like to build their companies and this is where he drops his first knowledge bomb Oh for that it will mean that 95% of what marketers use agencies strategist creative Professionals for today will easily nearly instantly and at almost no cost be handled by the AI and the AI will likely be able to test the creative against real or synthetic customer focus groups for predicting results and optimizing again all free instant and nearly perfect images video campaign ideas no problem I think what we are seeing here is something that you know it's an important because it shows us that when the technology is evolving it's going to be changing things in ways that we didn't think like even as someone that knows a decent bit amount of marketing and cative agencies and stuff like that I wouldn't have even predicted that you know and of course you can make these predictions yourself right now but I wasn't even thinking that you know you could have synthetic customer focus groups and you could have um you know systems and you know just rapidly testing and optimizing you know groups against ad campaign so if you don't know what that means so let's say for example you have advertisement on YouTube you might see one in this video but if you see one that is for example just an image likely what will happen is you know different people will see different images because they want to test the advert and see how well they perform so they might upload four adverts and you know instead of just you know putting one advert for £1 they'll put four adverts and put £5 each on them or $25 and essentially what you could do instead of that instead of wasting your money you could just have a synthetic group test against an AI group okay and see which of the ad creative which image which advert which video you know the AI the most and then once you see which advert performs really well with the synthetic group you could then produce that on YouTube and then it's put to humans and I think that would be pretty crazy so I mean like I said AI is really going to impact things that I really didn't think about and I think that is a really cool thing and I mean I wouldn't be surprised if someone actually does make a company like that in the future and if it's really effective so as well what we need to uh look at as well because there are a lot of Industry predictions that were really fascinating because in this video what I wanted to do is I wanted to get a really good picture on where everyone stands and then kind of you know look at where the median prediction is where the most common prediction is as well because I want to see where is a crazy AGI prediction and what's too optimistic and what's too pessimistic and then we can look at what actually is realistic so I wouldn't say you can plan your life around it because I think it's going to take us by surprise but so you can actually gauge where things are moving so you can see s samman tweeted in 2022 the future can be so good that it's hard for any of us to imagine my basic take on this is that we have unlimited intelligence and energy and all that it will unlock so the risks also do grow to the upside and he additionally added to this by the way I think we will have these before the decade is out not deployed at scale but it's clear that these are both going to happen so you know I do think that what he's stating here is that before 2030 and in order for that timeline of course that is just I guess you could say you know 6 years away or just within 5 years around that time so that would be around 2030 which is you know not too far away I mean 5 years seem like a long time but it will

### [10:00](https://www.youtube.com/watch?v=WVs6PXKiKmg&t=600s) Segment 3 (10:00 - 15:00)

fly by especially if we're seeing a ramp up in you know different a AI systems GPT 5 GPT 6 Claude 4 CLA 5 if we're seeing Google Gemini 2 Gemini 3 Gemini 4 you know multimodal robots all of those kind of things happening before we do get AI I think those things are going to be really crazy like unlimited intelligence and energy I think those two like those two things there are really insane so I mean you know this is the reason I've showed you guys this tweet is because this tweet from 2022 you know I saw this from a comment on Reddit that actually shows that you know it's still the same time frame as this recent statement where he's like you know 5 years give or take maybe slightly longer but nobody knows um and I think what's also important that says nobody knows exactly when or what it will mean for society and that's a really important thing because um as some people would put it is called The Singularity because nobody knows what's coming after um and that's of course you know reference to the black hole thing where like you know if you go in a black hole nobody knows what's in there so um yeah I guess you could say that you know Sam elman's prediction for AGI is around 2030 is I guess and I'm going to show you guys some other predictions by other AI figures that kind of gauge where the market thinks AI is going to be now of course I will talk about the elephant in the room of AI being achieved internally but I will discuss that later so bear with me for that moment so one person that actually did predict AGI in two years is Dario amod so this is the CEO of anthropic if you don't know Claude 3 so if you know Claude 3 Opus that you've been using if you've been enjoying it this guy is the CEO and recently on a podcast 7 months ago he said that AGI could be here within 2 years he basically said that 2024 is going to be a pretty stale year not stale but like pretty basic year but then 2025 is when things are going to ramp up in terms of deployment and in terms of Rapid Innovation and I wanted to include that here because this is the CEO of a major aiab that's finally starting to gain some you know widespread user adoption and I think his prediction is rather important on the market because he says the AGI is possible within 2 years which is you know I put the date up here which is the 8th of August 2025 I think that is going to be rather fascinating if it does happen because that would be a start contrast to you know samman's prediction of 2030 so you know that's a 5year difference give or take and I'm guessing you know I don't know the entire clip I was really trying to make this video as quickly as possible I might have added the clip in but I do think like I said before even if AGI is achieved I'm not sure if it will be deployed at that date so um in 2 years that is definitely a really fast date I mean is it going to happen I have not idea but then again we don't know what his company has in store for us because of course they're a private a company they're going to work on models they're going to you know mainly focus on safety that what that's what his company does focus on anthropic is a really Safe Company like their models didn't even really answer anything until the recent update and um they're doing so because they realize the inherent risk so um I'm guessing that you know within two years is it possible yes it's possible but um it will be interesting to see if this prediction does hold up now before we do get into some of the other predictions I think what is also interesting as well is that samman's um timeline has actually not changed and I think this is really important because when I show you previous people's predictions and how they've changed over time I think it goes to show what a Sam would know that we don't because he said in 2021 in 10 years I think we'll have chat Bots that work for an expert in any domain that you'd like so you'll be able to ask an expert doctor an expert teacher an expert lawyer whatever you need to have those systems and go accomplish things for you so I need a contract that says this I need to diagnosis for this problem I need you to book me this flight I want a movie created I want you to make me an anim short a photo realistic short that looks like this help me write this computer program so let's say the most repetitive human work and some creative human work you'll be able to ask an AI to do for you that's a massively transformative thing and I think the fact that he said in 10 years in 2021 and that's still at 2030 and now his timeline's only gone down by one I mean he's got a very good gauge on where things are headed and like I said his prediction is rather important so I think 2030 is going to be an important year now some other people in the community have said within 7 months which would put it at September 2024 which is this year which means that we only have around 6 months left to go um I kind of disagree with this just because I think you know AGI even if they have it I don't think they deploy it rapidly because I think it's going to take a lot of compute with the whole 7 trillion thing um and I do think that you know the only companies that could possibly do it are opening eye Google and anthropic so you know knowing that you know what opening eyes timeline is with the Future model releases I think AI in 7 months is just a little bit unrealistic although AI is absolutely insane and we've seen that technology can rapidly improve and evolve without any kind of you know prediction being solid it can just be simply Shattered by a tweet from openai so we honestly have no idea but again the problem is the problem with all of these predictions is they don't always Define exactly what AGI is and I think that is the giant problem with this kind of prediction date is that AGI isn't exactly defined but with Sam alman's definition of you know an AGI system that can you know uh go ahead and Achieve Noel breakthroughs on its own I would agree that 2030 actually genuinely makes sense and I'll get into why exactly that does make sense now someone

### [15:00](https://www.youtube.com/watch?v=WVs6PXKiKmg&t=900s) Segment 4 (15:00 - 20:00)

else who did Al AI timeline someone that you're gonna want to know is of course Elon Musk now like I said I wasn't able to find this exact sentence I was actually scrolling through this video a lot um and essentially in this video he said you know five or six years and I think that was in 2023 so that pulls us to about you know 2027 2028 that's when Elon Musk thinks we'll achieve AGI artificial general intelligence um you can watch that video I'll leave a link to that but um it's this pretty solid prediction Elon Musk has a very good trap record of you know and he you know the thing is musk has been you know stating all this stuff ages ago when if there wasn't much AI hype um he's been talking about the risk of intelligence for quite some time so it's important to note his prediction as well in addition to Elon Musk we also have Ray kwell who said that AI is going to be there by 2029 which is a very interesting date because you know apparently according to several sources 86% of Ray Coal's predictions have come true so 2029 being the year that also is quite similar to Sam alman's as you can see um that's 2030 here is 202 9 and I wouldn't be surprised if 2029 it does happen now I'm going to show you guys some other things as well in addition I'm going to talk about what I think makes sense now in this tweet I posted this on Twitter a couple of days ago and the reason I did tweet this is because this combined with the trademarks I saw combined with um samman and Greg Brockman stating that we're going to release these AI systems very slowly and combined with a document that uh you know looks a whole bunch of different facts I think this timeline does make sense completely although I would say that maybe you know could say 2028 for AGI um you can see 2023 gbt 4 2024 gbt 5 which is this year next year gbt 6 which is going to be AI agents music and video with Sora then of course 2026 which is gpt7 AI agents music video and embodied AI which is uh you know the robots and stuff and then 2027 we get AGI which is an artificial gener intelligence system um and I think it will and I think it's likely to be an embodied AI system as well but I do think we get embodied AI before that but I do think that this is going to be the year where things really do take off so that is kind of my prediction and I think it's a rather quick prediction in terms of you know where we're going to be but I think this one does make sense considering the fact that open AI said that they want to solve super alignment within 4 years so that's why I said that this dat actually does make sense because they did this in I think it was 2023 and if you add four years 2023 you get 2027 and we know that you know the date that they solve uh super alignment isn't going to be the date that they just go ahead and release AGI of course there's many different things that could delay and I will talk about those but the point is that I think if they're trying to solve it within 4 years then I guess that they you know they have a date in mind and they're um you know trying to think about you know how they could solve this problem before they realized that those systems are going to be built now um in addition I do want to say that the elephant in the room that of course you know while samman has said you know 5 years give or take there has been a lot of information to suggest that AGI has been achieved internally as with samman stating that himself and then just you know backtracking and saying I was only joking now I do think that they do have the path to AGI as in they know how to get there but they don't actually have the system itself but they do know um with the right compute with the right certain know scaling laws I think they do know how to get there possibly and I think there's a strong possibility on that will they actually release AGI you know as early as now uh I don't think it AGI is going to come anytime soon because I think they realize how crazy it will be to the public um releasing that kind of system so I think even if they do have AGI by you know the date that I've predicted combined with the document which I covered in a minute video I think that they will try and you know release it as slowly as possible just because of the ramifications that it will have um on society and that's not just something that I'm predicting that's something that Sam Alman and Greg Brockman have said themselves now in terms of other people predicting AGI Jensen hang the CEO of Nvidia has said that his timeline is 5 years give or take which once again brings us to the 2029/20 timeline of AGI being here now AGI the question on AGI is what's the definition yeah in fact that's kind of the Supreme question now if you ask me uh if you say Genson uh AGI is a list of tests you need to have a specification and you need to know what the definition of success is you need to have a test now if I gave uh an AI a lot of math tests and reasoning tests and history tests and biology tests and medical exams and bar exams and you name it SATs and mcats and every single test you can possibly imagine you make that list of tests and you put it in front of the computer science Industry I'm guessing in 5 years's time we'll do well on every single one of them and so if your definition of AGI is that it passes human tests y then I will tell you five years so like I said everyone has a different definition of AGI Sam alman's one is one that can achieve novel scientific breakthroughs and I think what the important thing as well is when I look at these kinds of different predictions in AGI I think Sam mman stating that you know we can

### [20:00](https://www.youtube.com/watch?v=WVs6PXKiKmg&t=1200s) Segment 5 (20:00 - 25:00)

achieve novel scientific breakthroughs means that his kind of AGI he's thinking of one that is probably virtuoso AGI so if you looked at the uh Google deep mindes levels of AGI paper they talk about three different kinds of AGI early AGI which is what we have now with gbt 4 um you know medium AGI which can complete 50% of the tasks as good as a human um we haven't got that yet and then virtuoso AGI which is 99% of all uh humans and then 100% a AGI which would be superum Ai and just completely dominates humans in every field and excels it in every single realm so I think you know samman here is stating that you know in terms of the level of AGI that we'll have you know by 5 years I think we could have an AGI that achieved scientific breakthroughs by 2030 that would mean that potentially we could get like a lighter AGI system even before then with gpt7 and GPT 6 perhaps so I think that is an important cavat to add because the definitions here are very important because whilst AG guy is a very hotly topic I think is important to try and Define that now one thing that I did actually find rather fascinating was I was actually looking at metaculus predictions and I'm going to switch uh tabs in a moment but you can see here that the community prediction is around 2031 now I'm going to switch tabs CU I really want to show you guys something I think one of the craziest things about this is uh you can see that the predictions are pretty crazy okay because if we take a look at the uh prediction you can see that in 2020 I'm actually going to zoom in is quite hard to see um but you can see at 2020 the community prediction was uh I think it was you can see right here it's 2047 so just look here look right here as I scroll see right here it's 2020 and then 2047 so people are saying that 2047 then as you go as you go the predictions kind of even get up like we went from 2047 to 2057 so people added 10 years to the AGI deadline and then you can see as soon as chat gbt was released boom we went down to 2035 which is pretty crazy and then you can see as we continue then GPT 4 gets released and then the date comes down again to 2031 232 so the community prediction on metaculus is around 2031 2032 and you can see the kind of forecast that we're getting and every time like I said every time there's a new AI system released of course those predictions are now of course these predictions can change as they do but um another prediction as well that you're going to want to see is of course this one right here so uh these guys use some economic modeling techniques to predict the expected Ral date of AGI and they said that the average predicted date from this analysis is 2041 with a likely range of 2032 to 2048 and an estimated earliest possible arrival date of 2028 which is just 5 years away so 5 years away seems to be the key Zone like from 2028 to around 2030 that time zone seems to be the time where we're going to get you know virtuoso AGI and then of course you can see here um one thing that they actually do talk about which is rather interesting they talk about significant global scale conflict in the interm so one thing that they actually do talk about and I think this is rather important to talk about because uh I think around you know it's every 10 or 20 years we do get a Black Swan event and that Black Swan event it does take the entire world by storm the previous Black Swan event that we did have was Co and of course one before that was the uh you know financial crash in 2008 so these kind of things can slow down the economy can bring the economy to a ho and can add a couple of years to the prediction as things do you know increase and as things do get better we do know that certain predictions do increase on the timeline with certain breakthroughs but you know certain things also do delay them now um they're stating that this wouldn't add more than 10 years but I think one thing that people haven't talked about is Will policy slow it down so recently you know literally just today as I was reading this I came across an article that said you know exclusive the United States must move decisively to overt extension level threat from AI government commissioned report says so I'm wondering I am wondering okay is this and I'm going to go on this article in a moment is this going to be something that happens to open a I think because open has such a uh prominent role in AI development I'm wondering I keep wondering are they going to change and just like you know is the government going to I wouldn't say infiltrate open AI but are they going to somehow take control because like I said if they are developing what is known to be as the most powerful most uh you know advanced technology you know that's stronger than nukes and you can see right here they've got the nuke images um if they've got something that's you know arguably bigger than nukes how is the government not going to intervene and say look we need to know what you're doing we need to make sure you're deploying this correctly I think at some point at some timeline somewhere down the line the government slows things down because I just think you know it doesn't really make sense that you know a compan is going to have something more powerful than nukes and they don't get involved somehow so that would also be my prediction on something that could potentially sew something Downs you can see the article says you know the US government must move uh quickly and decisively to avert substantial National Security risks stemming from AI which could in worst case cause an exension level threat to the human species and it says current Frontier AA development poses ENT and growing risk to National Security the risks of advanced Ai and

### [25:00](https://www.youtube.com/watch?v=WVs6PXKiKmg&t=1500s) Segment 6 (25:00 - 27:00)

AGI has the potential to destabilizer Global Security ways reminiscent of the introduction of weapons and AGI is a hypothetical technology that could perform most tasks at or above the level of a human and such systems do not currently exist but the leading a labs are working on them and many AI AGI and many expect AGI to arrive within the next 5 years or less and what's interesting is that you know the report recommends should be set by a new federal AI agency although the report suggests as an example that the agency could set it just above the levels of computing power used to train Cutting Edge models like open eyes gp4 and Google's Gemini you can see right here that it also states that the new AI agency should require AI companies on the frontier of the industry to obtain government permission to train and deploy new models above a certain lower threshold the report adds authorities should also urgently consider outling the publication of weights or inner workings of powerful a models for example under open source licenses with violations possibly punishable by jail time the report says the government shouldn't further tighten controls on the manufacturer and export of AI chips and channel Federal funding towards alignment research that seeks to make AI safer so this is something pretty crazy because I think it was India that actually recently had a law which talked about how you need to get permission before training AI models or something like that and many people were you know rightly so speculating that this is going to slow AI development down a lot um and I'm kind of interested to see how the political landscape actually plays out with such a fast moving techn ology that is AI because it seems that they're never really going to catch up so I have no idea what the us is going to do it will be interesting to see if they actually do implement this stuff because on one side you do want the safety of the public to be protected but of course on the other side you do want the US to be the leader in AI if you are someone that lives in the USA so it will be interesting to see how that comes but what is your prediction on AGI do you think it's coming within 7 months 2 years I think it's probably going to be here by 2028 I think of AGI system could be there at this time I do think that by that time by that prediction I do think there will be some kind of government intervention if they start to realize how crazy these Technologies are especially if they start to you know really have those agentic capabilities so with that being said I'd love to know your prediction what you guys' thoughts are and there was anything missed in this video leave it down in the comment section below

---
*Источник: https://ekstraktznaniy.ru/video/14475*