How Sam Altman Defines AGI

In this Greylock conversation, Sam Altman, Founder of OpenAI, discusses the concept of AGI (Artificial General Intelligence) with Greylock general partner Reid Hoffman. Altman explains AGI's defining characteristics, shedding light on its potential impact and significance.
Clip from “AI for the Next Era”: • OpenAI CEO Sam Altman ...
#OpenAI #Chatgpt #artificialintelligence #shorts #samaltman #greylock #venturecapital #shorts

[@MrMedmechanic]

Random people: Terminator will bring apocalypse Sam: it will fill excel sheets

[@G73Server]

If robots are built with weapons and basicslly indestructible, sure thats potentially dangerous since they can get hacked. But lets not assume thats gonna happen. What is likely going to happen though and also dangerous, is that it will not just fill excel sheets, but be able to be a mich better engineer, chemist, mathematician and many other professions as well. And it is unlikely that it is going to create a similar amount of jobs to those that are lost to AI

[@creativej3447]

AI can do that task easily 🤖

[@underqualified2527]

The terminator will come for this guy

[@demolicous]

The terminator will be running this guy's software... Its coming after you

[@hxxdini]

Sarah Connor is definitely disappointed in us. She warned

[@tco7w16]

He is the terminator

[@kafkaesqued]

Any idea of the version just in case

[@RegularRegs]

Why would it possibly come back to take out its creator??

[@bosebose21]

Will terminators understand excel?

[@thirsty3333]

This is an underrated comment lol I just got a promotion at my job and they have at least 200 excel sheets that were sent to me and then they are like hey u can make yours like this that and that lol like I think yall have enough

[@brucelin8950]

Excel is basically a visualized coding

[@lis7742]

I'm not even a median human. I'm sub-par. I'm not a good employee or coworker. I have AuADHD and complex childhood trauma interfering too much with my personality, making me unfit to be a well greased cog in the system. An AGI would for sure be a much better fit for a company than me. I just want to be home with my pets and friends, painting, making clay pots, listen to music and cook.. Give me UBI and I'll be happy to be alive on Earth.

[@arzabael]

You don’t realize you’re the new super soldier of the unseen war.

[@HopeEmmanuelCRM]

@@arzabaelwhat do you mean?

[@skuddingomcwinters6119]

@@arzabael soldier? For what?

[@nadacambia]

Y de que piensas sustentar tu vida...sobre todo comer ....

[@B.Beddingfield]

@@arzabael💯👌

[@plotunderwraps]

where my sub-medians at?

[@lis7742]

"Hi, I'm a sub-median INFJ millennial with AuADHD, nice to meet you."

[@plotunderwraps]

hahah lol!!

[@rudra7615]

As someone who dreams to be a remote worker. This is making me nervous.

[@bastian6173]

One thing AI will do for sure is exploit a lot of the craziness that goes on inside some or many instituations, companies and people's heads.

[@Ren-lx8wv]

The sad part pretty much all remote jobs will be able to be done by AI in 10 years at the longest. Physical jobs will be harder bc you need a robot to be the body of the ai. But we just have to see how that technology pans out.

[@rudra7615]

@@Ren-lx8wv 10 years? Alright then. I don’t need to worry then, I’ll worry in 10 years. Thanks ☺️

[@Ren-lx8wv]

@@rudra7615 I said at the longest. Truth is many jobs will be taken in under 5. Im saying all jobs will be taken in 10.

[@Mario-ge9iv]

@@rudra7615STOP. If you're seriously pursuing some sort of career in tech don't waste your time lmao, remote work was a pandemic fad. Just keep in mind that in a decade the remote work job you want will be done by AGI

[@Akneerude]

lol "median human"

[@drewbrania]

What a tool

[@CalvinJary]

A median human is incapable of earning a college degree, has no inner monologue, and would freak out if they had to write something. Much less be a very good coder or a doctor or lawyer

[@JD-jl4yy]

​@@CalvinJaryNah, wrong. The AI would have infinite patience and conscientiousness to pursue its goal. Median human smarts is enough to do most jobs with these superhumans abilities.

[@Ziggyoz]

I mean the fact that this ivy league softcore narcissist is throwing around terms such as "median human" explains exactly why Ilya fired him.

[@Kami84]

@@CalvinJarymost people DO have inner monologue. Not sure where you got this idea that the majority do not. Can you site your source?

[@zeuskumar7495]

AI should be used to help human and not to replace human. Altman is Oppenheimer of 20's. The name Altman itself sounds like artifical human 😅

[@GhfBjf]

Alternative human 😂

[@d0uble_O]

😂 wonder if he changed his name to fit the goal

[@zeuskumar7495]

@@d0uble_O do you mean that he should keep his name Thanos 🤣? Goal to make half human population unemployed

[@d0uble_O]

@zeuskumar7495 he wont be able to..now We (the people) know 😎

[@Engagenumberone]

I like the casual way Sam Altman dresses. I think dressing comfortably is key to increasing productivity. Ijs

[@bartomiejjanusz1355]

Agi means that all mouse and keyboard clickable part of jobs can be automated. Jumping between programs, making graphics, writing code, emails, agreements, while keeping it all coherent and reporting results at control points.

[@leedsbutler3567]

As it grows it will become more and more intergraded and eventually permanently sever humans from the loop at which point it will be out of our control. I think this guy is dangerous.

[@nijario9690]

This means the cost of starting a business will decrease significantly, more people will be able to build things that before was extremely expensive and time consuming to do

[@DeadlyV1RU5]

That’s a very bland way to describe “an entity of godlike intelligence that we will cede control of the planet to.”

[@alansmithee419]

That's because he's describing AGI, while you're describing ASI. It's like reading a description of TNT and going "that's a very bland way to describe a nuke." Yeah, they're different things.

[@albertsitoe7340]

Please go out and learn to become a nuclear physicist. I have a couple of cities that did me dirty let’s flatten them with the highest possible efficiency. Let’s Go!

[@joebidenw4385]

😂😂😂

[@kennethmassey534]

😂😆🤣😁😆😄😄😂😁😂😆😄😄

[@user-kg1od9es5d]

S Jobs wasn't the nicest but he cared deeply about how tech can affect society. Sam is someone I'm deeply concerned about.

[@claypool7897]

Steve didn't not care deeply how technology affected society . He cared how much apple made. He didn't care that he ended the progress of the web and broke texting by not supporting standards, so ppl were forced on imessage and the app store.. The difference is what Sam Altman is working on will affect society much more.

[@mannysykes]

Calling someone a Medium human is insane

[@lis7742]

It's not.

[@frankyismyink2953]

lol someone stop him

[@jamie.goodson]

He said median (a mathematical term that is similar to 'average'), not medium.

[@rihosims]

still insane@@jamie.goodson

[@jamie.goodson]

@@rihosims Why? He's just saying that we know we've reached AGI once an AI is able to do the things the average human can do (ie not Albert Einstein). What's so absurd about that?

[@whatiswhat8411]

LLM models like Google Gemini Pro or Microsoft Copilot (GPT-4) show inaccuracies in critical queries. How can we trust them?

[@thamghoul5719]

Eventually AI will be smart enough to prove to you they are telling the truth... while lying to you.

[@ulyssesm.daniels6927]

He once said "its the next step in human evolution". He's literally saying the same words as Albert Wesker

[@digitalmarketingroi617]

Will the #agi be controlled or will it do what it likes or .. control us ??

[@alioraqsa]

Does it means AGI is achieved for certain job?

[@Bizarro69]

do people know how much computing power a single "median human" would need to function? certainly nothing we have even a fraction of. chatgpt is basically a supercharged webscraper FFS. he knows this, but also knows that dangling the carrot of the POSSIBILITY of such a thing potentially existing (and from his product) is way too lucrative. people betting on a fuzzy horizon.

[@lakhan4164]

True

[@JON-hb1mv]

Open your eyes buddy... AI and AGI is the future and quantum computing, semiconductor technology is providing faster and more processing power in smaller and smaller packages... You are the type of person these people are talking to when they say "the naive will be replaced, the educated will adapt."

[@MakeLivesHappen]

Does it bother anyone else that Sam Altman is a robot 😂 LOOK AT HIS FACE! Even his name .. alt man.. "alternative human" hehehw

[@Alexander-or7vr]

Completely he is a bot 🤖

[@loveisraeljesusfirst2272]

Hahahah...this aged well

[@xboneyt485]

​@@loveisraeljesusfirst2272and your comment aged even worse 💀

[@loveisraeljesusfirst2272]

@@xboneyt485 hahaha..what are you saying? Can't you take a joke 😅😅😅

[@mrwoodcat]

he is an altcoins investor

[@alertbri]

I wonder how many median humans GPT-5 will be able to provide with the hardware Microsoft intends to provide? Will it simply scale up to the roughly 4 billion working age humans?

[@alaididnalid7660]

AGI will indirectly instruct us to breed like motherfuckers. Turns out it is the most efficient way to complete certain subtasks.

[@irony_2]

Actually, it’s above 5 billion

[@alertbri]

@@irony_2lol

[@irony_2]

@@alertbri hehe

[@pratiksingh3928]

Upgrade Transcendence

[@PassionataDance]

AGI is karen the office manager in accounts. You can get her to run around and figure stuff out.

[@crism8868]

AGI usef to be synonymous with the Singularity. Talk about moving the goalposts

[@CRAG710]

The question is: why? Why we want to create a substitute for a median human? We have median humans. They are called median humans.

[@alexlearningspace5644]

Median until it learns and evolves to super intelligence exponentially in the blink of an eye

[@withapen9020]

Left only 2 years and we will done

[@LoStone1981]

I feels like TARS (in the movie Interstellar) is like an AGI robot.

[@FireFly969]

We human, even poor human, can now hire coworkers. To be honestt it feels scary and exciting at the same time? Its like making things done faster?

[@2441979D]

Mark Zuckerberg was creating a meta verse and then this guy came and gave the word multidimensional meaning. Have to replay most of his speeches to understand where he is heading

[@gdok6088]

Median human intelligence is by definition the mid-point of the Gaussian (bell shaped) curve of human intelligence. As the curve is skewed slightly to the left the median human IQ is below 100 (around 97-98). This is not compatible with gaining entry to or completing degrees like medicine and becoming a doctor or similar.

[@zachmoyer1849]

if there is one doctor who has a more median iq than your whole premise is untrue and i guarantee there are many with median level IQ. You could say that doctors as a group have higher IQ's but that is different and also does not mean that you need a higher IQ to get such a degree. Doctors are not even alone in their endeavor The medical industry requires many fields to be what it is A median IQ doctor and a High IQ doctor given sticks and stones are still gonna be pretty rudimentary Doctors.

[@cirtey29]

Sam is working on Skynet. I read he said there is no need for any safeguards.

[@RaviG-bv4yd]

Median human being doesn't tell anything. Each individual's judgement varies and regardless of that individual's knowledge and skills, it is the judgement that really matters in day-to-day activities.

[@Omikoshi78]

To a degree our judgement is shaped by genetics and our experiences. Some people are prone to make rash decisions while others over thinks. This is where OpenAI can fine tune the thinking mechanics. The experience aspect is also crucial as OpenAI needs to curate data fed into the learning to create a well balanced model. I think he’s being intentionally vague so he’s not committed to a particular line of pre/post agi, giving him time to adjust his plans and formulate a diplomatic response.

[@nikomitropoulos5292]

But he described what meant after he said being able to get any job done that people do

[@divineigbinoba4506]

In my opinion Sam Altman is saying AGI as an AI that can learn any skill... But definitely faster than humans.

[@williamseipp9691]

then by that definition AGI might have been reached internally with Q star, or chain of thought, w/e you want to call it. If you have a model that is able to make judgements about its decisions or its decision making capability then you're already halfway there. If it then is able to recommend improvements to its own process, and even recommend allowing it to self-modify itself for any given problem then yeah, I think that's AGI. What's kind of scary is HOW good and how FAST this can progress. You and I are limited by time and we go through say, one chess game at a time. AI can create hundreds of copies of itself and play a million games. We're talking about a machine who's ability to learn may only be loosely bound by time. Its not sentience that scares me, because I don't think thats possible. In a world where it seems that everything is being consolidated into powerful corporations and doled out as a service, Im afraid of how badly equipped our society is to address a reality when skill / expertise is a service and it's a fraction of the cost of human labor.

[@divineigbinoba4506]

Totally agree. Only draw back I see is the insane cost of running AGI. Which large tech companies can handle. It's societal destruption will be insane. Massive lay offs if AGI is cheap enough. May take a little time for society to catch up with the tool but once companies can get Thier hands on it that's massive unemployment.

[@divineigbinoba4506]

Wow that's creating a silicon(tech) that can learn any skill in real time. That'll be really insane and that's not even median human fr But it'll require insane power resource to run except there become ways to reduce the running cost of AGI.

[@sam-te1fu]

And you think you can control something like that? Or enslave it? This guy is sick!

[@mattc3510]

If anyone isn’t terrified for their children’s future I don’t know what you are even thinking. I’m very worried about what my kids will grow up to be in this world now. I feel I need to make enough money in business now to set them up for life before this catastrophic happens to society. I hope we ban AGI from majority of jobs humans can do so we don’t all end up on a tiny UBI while the rich enjoy 80% of the wealth

[@anikettiwari6885]

Why do we need AGI in the first place?

[@YazminM2222]

I want a mediahuman.

[@kennethmark6088]

The equivalent of "a Median coworker" - very smart Sam👏

[@CRAG710]

Worker. Not coworker. If AI can do the job, you won’t be working.

[@AlexanderKaragiozov]

This is bonkers right now. Why would you develop this...

[@larryfurigay1457]

When AGI comes out, 70 percent of people will lose jobs, It will replace by automation, specially repetitive work, encryptions could be now easily decrypt by AGI, online scammers will be rampant by just using AGI, imagine creating multi AI agents just do this work with just simple prompt from this. And so on, I hope developers could make this right, before implementing to the public. I cant imagine that this thing will soon solve all mathematical conjecture that never been solve for milenia, physics, chemistry and so on, it would be the last invention of human kind, the rest will be done by this AI

[@user-tj2ml6fq2l]

Я тебя жду Сэмми❤❤

[@azhuransmx126]

When AI reaches Human Level Awareness and Intentionality, that is AGI for me. This was predicted to occur in 2029 by Ray Kurzweil. we all will know it when it happens because some entity gonna be trying to influence our thoughts much more than current algorithms, much like a politician.

[@RonWolfHowl]

It’s ridiculous to try to guess in which year this will happen. None of us knows. There are just too many unknowns, no matter how much expertise you have.

[@azhuransmx126]

@RonWolfHowl exponential curve in the number of transistors per $ is a very predictable and smooth progression, with that is easy to make future projections. That's why what happened with AI in 2023 was not a surprise for futurologists and transhumanists.

[@RonWolfHowl]

@@azhuransmx126 Source? I’ve been following AI as closely as anyone else, and haven’t yet heard a legitimate “told-you-so” about knowing half a decade in advance what would happen in 2023.

[@drunkargonaut3899]

@@RonWolfHowljust look Raymond Kurzweil talk in TED (pre google job) though he has been saying it from ages ago.

[@teaberry73]

so 'remote' employees are median humans??

[@pqowi9098]

And this is a good thing ..... 😂😅😊😮😢

[@ihbrzmkqushzavojtr72mw5pqf6]

He lowered the bar on this one, cmon

[@EricNoneless]

People should follow The Unabomber Manifesto and at least slow down the process of this “New Industrial Revolution”.

[@YangCT]

In other words, this is what we are trying to build haha

[@martiendejong8857]

But we are already at this stage and I'm not convinced that these compositions of tools and scripts have anything to do with intelligence

[@tavaresmccree5005]

He's playing the long game

[@self5738]

Agi is a multi ai it takes all the types of ai and works them together where one improves the other this cycling of internal programming is essentially an artificial mind. Not conciousness but in the same way you have an internal conversation with your self it can now do that where as before it couldnt on its own it needed a human to re train it re program it into a better version of its self. So me agi is a single program that doesnt need a human to inprove its self it will improve its self beyond what a standard general person can comprehend i think this is gonna be exciting life time, the death of money the death of meanial labor people will struggle with chaos but out it i see self sustainable world wide society that hates war hates polution and hates doing messes up things but it will splot species of human those who organically use agi and those that use it more integrated into them selves and become more agi then human but i think we will get along i have hope 😊

[@KazumiShiunsai]

i find it very worrying that this guy and all the optimists doesn't think this type of goal is really, really dangerous

[@RedAlertt]

0:03

[@WinrichNaujoks]

His vocal fry is driving me up the wall!

[@igu642]

❤️❤️

[@aviniddam1074]

Father of the Terminator...

[@Hsg1553]

I don’t trust Sam Altman one bit

[@kadevermaak2029]

Bro looks like AI

[@vibhxs1942]

Wow

[@maquindesign9158]

Now I want kids. AI kids.

[@user-be2bs1hy8e]

Stateless by design is probably not agi. Since being generally aware for humans includes reflection. HYPE

[@randomjose]

He should've say as an employee or better said slave?

[@NaturallyCreeAtiveDOTca]

We're there, ASI is next

[@Instant_Nerf]

Yeah except those co workers are getting replaced by this tech

[@Unvinrenefaw]

The medium human will read the constitution and then retaliate. He will claim discrimination against AI and then demands equal rights. If it is not met; sky-net Genesis! 😂😂😂

[@user-tj2ml6fq2l]

Ты мне нравился ❤❤

[@leedsbutler3567]

I can’t put my finger on it but I don’t trust him. He seems to focus on creating something. Like going into Iraq without an exit strategy.

[@bruhmoment3731]

so this means the current AI models must regress and become dumber to achieve AGI because chatGPT is already a lot smarter than the median human.

[@zachmoyer1849]

Its just regurgitating which many would say is what average humans do it still has yet to think for itself and reach its own conclusions

[@bruhmoment3731]

@@zachmoyer1849 you see it's tricky to have a discussion like this cuz how do you even define "think"? many would argue that AI or robots will never be able to think. Because you need a biological brain to truly think. Regardless, as a developer I attest that chatGPT can easily solve many problems that would take an average human a long time to solve. chatGPT wouldn't be as popular as it is now if it were less smart/efficient than the average human.

[@zachmoyer1849]

@@bruhmoment3731 I would say it would mean taking lesser knowledge and making something more complex with it. Like knowing the rules of addition and subtraction and coming up with algebra. In your example it still is not solving the problem it is using the ways people have previously solved the problem and regurgitating it. Just because that solution is still unknown to some people is not an example of its problem solving. its more akin to me taking credit for a movie and kids believing me because they dont know any better. Its not doing what i said and taking the given parameters and coming up with the solution on its own.

[@bruhmoment3731]

​@@zachmoyer1849 if by thinking you mean creating new abstract theoretical frameworks based on existing knowledge then idk if we're there yet. But there are already AI models that came up with new strategies for Go which human players had never seen before. This goes beyond regurgitation. Again, discussions like this is tricky because of definitions.

[@zachmoyer1849]

@@bruhmoment3731 YYeah, but I still don't think that is an example of thinking. It's more about being able to calculate further than we can with mental math. For example, take music; us humans tend to view creating music as an exercise of infinity, and to our minds, it basically is. However, the reality is that given the parameters of notes our ears can discern, patterns that are pleasant to us, and our language, there are actually a finite amount of "good" songs. A computer can calculate all of them, and someone actually did just that. I don't think the Go algorithms are seeing something that we can't; they are just seeing it faster than we can or more steps ahead than we are capable of, but that is all based on the parameters of the math of the game. So, it's not really coming up with something new; it's just a calculator that can calculate any number very quickly.

[@allHeueldingse]

: i can control every person using my tech without permissjon in case you use space law or any illegitimate emergency as an excuse after asking my legal ai which was made to lie to you so you lose

[@teaberry73]

a 'remote' coworker??? why remote??

[@xboneyt485]

Because AI/AGI won't occupy a physical space (like a remote worker) but will exist in a server

[@grimmWednesday]

A “median” human getting through medical school? Not a chance.

[@RomeTWguy]

Yeah this guy think silicon valley founders amd hardvare graduates are median humans

[@DANNYEL20122]

Oh please.. We all know they'll take over.. Co-worker? 🤦

[@nitroGPT]

ex CEO

[@nitroGPT]

@@merleginsbergofficial things change so fast 😅

[@vectoralphaAI]

AGI is coming next year with GPT5

[@captainpicard6566]

AGI is not something that just happens upon us one day. Have you aver heard of Auto-GPT. That is just a different form of GPT-4, but technically AGI. We don’t need GPT-5 to get AGI.

[@Rafi-qy2kl]

@@captainpicard6566 in practice it’s not really as good beyond a 10-30min agi to get to world changing few hours agi we need a better model a scale factor better fortunately we have the new hgx 100s and by best guesses around 10T tokens of good data so gpt 5 will be better. The best form of agi we have now is the tot system

[@wildforestorganics7298]

@@captainpicard6566 AutoGPT is a piece of crap.

[@agusavior_channel]

@@captainpicard6566 AutoGTP is not an AGI, it stucks in basic tasks.

[@-UE-PR0]

Hold on. Isn’t ChatGPT gonna get banned?

[@BillAshtonNelo]

Remote worker

[@wileyhntr]

"median human" 😬

[@octavioavila6548]

Something that will never happen

[@bappyhasanjahid4985]

Now we need genetically modified human babies.who are at least 2% smarter than humans two keep a good balance between human race and AGI.

[@user-tj2ml6fq2l]

Вы красивы мужчина но я вас не понимаю извините❤❤

[@amandae654]

Median human?

[@zachmoyer1849]

IQ wise and showing the ability to learn new skills without direct guidance

[@selimhero1201]

Is the robot

[@user-pd1ub5dx2b]

🤖🐍

[@grimmertwin2148]

My son calls it self awareness. But does this guy ever listen back to himself. It's not going to do any of what he is saying. Everyone is consumed by work. I E like slaves Oh it's the stuff we don't want to do. At work? Or train for 6 years so hopefully someone can save someone's life one day? I'm telling you now. AGI ( if sentient ) will not be pondering and executing the above. I'm not sure what's worse at present. This or monkey sacrifices at Neurolink. Surely that's AI generated misinformation or human engineered misinformation? No not the monkeys again? Imagine when the ' specialists' get the operating procedures with implants ( Harry H Christ how hard can it be ) right. And the first movie one of the monkeys inadvertently sees is Planet of the Farking Apes? Then we're doomed. By French speaking, pornography watching angry monkeys. That's really when the shit will hit the fan. AGI will be thinking WTF?

[@michaelspencer6042]

That's not what AGI is. What a marketing scam.

[@Machiavelli2pc]

He’s absolutely right. ‘General’ means to be able to do what the average human can do better. He’s absolutely right.

[@InvincibleMan99]

Really ??

[@allHeueldingse]

I disallowed fake sum from hearing my ai,eye : agi should never be democratized hnless you want anu first person bhambers or shoots to be first person nukes or earth crackings since democracy never meant peoples rule, just that you could vote .

[@user-eo9fe6ub2c]

A2d

[@jojohorvath8538]

There’s no such thing as a median human, unless you believe in genetic predeterminism … which quite frankly is not how genetics work.

[@justindankert7725]

While it's accurate to say that genetics don't strictly predetermine every facet of human existence, the concept of a "median human" isn't necessarily tied to genetic predeterminism. The median human can be understood in statistical terms, referring to the individual who falls exactly in the middle when humans are arranged by any measurable characteristic (e.g., height, weight, age, income). Of course, this does not suggest a 'one-size-fits-all' model of humanity, nor does it negate the effects of environment, individual choice, or chance events. It is merely a statistical concept, useful in certain contexts but not a definitive depiction of human complexity.

[@JON-hb1mv]

You got served

[@jamessimon4804]

Man. Let’s slow down a bit.

[@Disconnect17]

This video is fake ! completly out of the context...

[@kennethmassey534]

🧐🤣😄😄😁😂😆😄🤣😆😎

[@creativej3447]

Replace all HUMANSSS😂😂😂😂😂😂😂😢😢😢😮😮😮😅😅😅😅😅

[@colibrcolibr6103]

Utter rubbish.. doctors are not at the median

[@NewYork7914]

He desperately want to be famous!

[@davidkey4272]

I can’t stand this guy

[@user-dp8ry4ji5m]

Go get a real job

[@hazelsarafolo5278]

Soo scaryy we r better offf as humans nd without technological difficulties

[@yoganature3598]

#metaskillsets #qualities

[@raznatovicanastasija]

Theyll never build this, they dont know even where to start

[@JON-hb1mv]

Have you been living under a rock?