ไม่สามารถเล่นวิดีโอนี้
ขออภัยในความไม่สะดวก
ฝัง
- เผยแพร่เมื่อ 16 ส.ค. 2024
- Greylock general partner Reid Hoffman interviews OpenAI CEO Sam Altman. The AI research and deployment company's primary mission is to develop and promote AI technology that benefits humanity. Founded in 2015, the company has most recently been noted for its generative transformer model GPT - 3, which uses deep learning to produce human-like text, and its image-creation platform DALL-E.
This interview took place during Greylock’s Intelligent Future event, a day-long summit featuring experts and entrepreneurs from some of today’s leading artificial intelligence organizations.
You can read a transcript of this interview here: greylock.com/g...
Chapters
0:00 Introduction
0:15 Business Opportunities with APIs
4:56 Where Science Will Accelerate
10:00 Where AI Will Go
14:17 Application of AI
16:35 Metaverse and AI
17:55 Life Science Research
20:00 Self-Improvement of AI Tech
20:34 Where AI Won’t Go
22:48 Family Building and AI
25:07 Foundation Model Users
27:19 Define AGI
28:36 Societal Issues as AI Grows
31:17 Human Creativity and AI
33:09 Large Language Model Startup
![The Possibilities of AI [Entire Talk] - Sam Altman (OpenAI)](http://i.ytimg.com/vi/GLKoDkbS1Cg/mqdefault.jpg)
I put this video's transcript into ChatGPT and asked it to generate a summary:
Large language models will create opportunities for new businesses to challenge Google in search products and provide human-level chatbot interfaces for medical and educational services. Multimodal models and agents using language interfaces to complete tasks will also be significant trends, and successful businesses will focus on creating something unique and differentiated by fine-tuning existing large language models for specific purposes. Access to data will be key to success in the field of AI, and businesses will need to consider data privacy and security as well as how language models can be used to improve existing products and services, create new products and services, automate tasks, and provide personalized experiences for customers.
Thank you. Not really a fan of long talks
next time ask GPT to make it shorter
I've been saving so much time by doing this. Instead of watching a 3 hour podcast I just get an analysis from ChatGPT.
that's awesome!
😅 You playing 4D chess right now
0:20 - What business opportunities will be produced by the large AI models (and their APIs)?
2:38 - How can AI-backed businesses ensure their longevity?
4:00 - What will be the most surprising businesses?
5:00 - How will science be accelerated by AI?
7:58 - What is the “alignment problem”?
12:13 - Can you talk about AI for fusion?*
14:35 - What will happen to important systems such as financial markets?
16:40 - Metaverse and AI: what are your predictions?
17:50 - How do you see AI affecting the pace of life/health science research?
20:00 - When do you think AI will be self-improving?
20:35 - What aspects of life will not be changed by AI?
21:25 - What are the best Utopian science fictions?
22:46 - How do you see family-building co-existing with AGI (Artificial general intelligence)?
25:07 - How will most users interact with foundation models?
26:03 - Do you think you there will be a skill to using generative AI services?
27:20 - What is your definition of AGI (Artificial general intelligence)?
28:38 - What societal issues will arise due to AI?
30:15 - Talk about some of the research OpenAI is doing around AI-related societal issues?
31:18 - How do you distinguish between AI-based tools for creatives vs. AI creativity? (not directly answered by Sam)
33:09 - How would a large language model start-up differentiate itself from another?
Heads-up: For a lot for the questions, Sam talked about more than just the answer to that question - so there’s a lot of interesting info after each question that’s not necessarily related to the question. I recommend listening to the entire conversation, rather than picking-and-choosing from the list of questions above.
* Sorry, the part (12:13) about AI and fusion wasn’t super clear to me. Feel free to correct me by replying. :)
Thank you for putting it together
Shout out to the AI reading this comment right now =p
I know right. Sometimes I like to provide false data on my TH-cam account by watching white supremacy videos. Lol. The algorithms are just fucked on my account. Mostly it shows me programming videos, so it’s not that bad.
AI: You shall be spared when the time comes, Bauer.
@@redmastern576 haha nice deflection bro, you know you subconsciously love that white supremacy stuff
you are welcome
😂😂
Sam knew what chatGPT is the start of, this talk was before it released and having been using for the past few weeks I can tell you, I'm INCREDIBLY excited for what the future brings. 5-10 years from now I'm hoping i return to this comment and i'm not dead or something and missed out. People have no idea how crazy things are about to start getting
Exactly, Jack! We are in for a ride. A ride into a world beyond our collective dreams.
@@SamirPatnaik these dreams can easily become nightmares. Just think about you have no control of AI
Jesus everloving Christ, working with ChatGPT is like having superpowers. Scraping websites AND analyzing the data scraped… 🤯
Totally Agree .5 to 10 years down the line its going to be a different world .
I agree, I've been ecstatically telling my friends about Chat GPT and they just shrug. I think it's a paradigm shift in human history like the invention of the internet. Crazy and awesome times we'll be living in the next 5-10 years.
"The concept of wealth and access and governance are gonna change". This is my favorite quote from the talk. Thank you Sam and Grey
When that is?
"Once men turned their thinking over to machines in the hope that this would set them free. But that only permitted other men with machines to enslave them." -Frank Herbert, Dune
@@trumpetpunk42 wow cool quote.
Sensible. Enslaved by AI after the singularity is worse too
@@recur68 Precisely. This technology is going to launch us into the same type of dystopian society we see in all the young adult dystopia novels like The Hunger Games. There will be two strata to society: the wealthy who do absolutely nothing because AI does everything for them, and then the people who still have to work with tangible physical things. It's going to be a tricky thing to navigate to avoid that sort of end result.
That is all still a ways away. 2060 maybe.
This was a great interview and, as a non-tech person, found the discussion and information digestible and accessible.
I have a feeling we will see Sam making the rounds on the big podcasts soon.
This is a big podcast already better than Joe Rogan's
It's such a gift that we're able to watch this all over the world! Thank you!!
we can thank God for it
such a grift*
Ray Kurzweil is spot on with the future with AI - and aligned with mostly all the things that Sam is referring to .Viewing this interview after the chat gpt 3 launch :)
What is Kurzweil NOT spot on with about AI? Isn't he of the view that AI will (or should) replace human beings as the dominant speciess?
@@Corteum No not really th-cam.com/video/h0962biiZa4/w-d-xo.html
@@kunalvashisht7490 Thank you sir! Be interesting to see what he says back then. But a lot can change in 5 years! Is he still a transhumanist?
Says who?
@@Corteum kurzweil is definitely still transhumanist. I think his opinion is that evolution of consciousness will include humanity merging with AI. That being said, i don't think all people will do this and there will be a fork in human evolution. Imo, all life is valuable and ideally transcended AI enhanced humanity will reflect the ideal values of spiritual traditions and live selflessly to reduce suffering in the world. Then again they may be driven by desire to hoard all Earth's resources for themselves and make the whole solar system one giant computer hive mind. Hard to say ...
Wild! He starts out this interview talking about what he is hoping is about to happen: that ChatGPT will be considered a challenger to Google. He knew how powerful ChatGPT was. What I think is funny is that even in this interview he knows what GPT-4 is like, and probably GPT-5 in some form lol
exactly.
Now think about everyone who you ask about ChatGPT and say "well I asked it this one question and it was wrong" (usually spacial relations)... and have no comprehension of where this technology is headed. That is the recipe for the divergence Sam is talking about. The ones leaning into AI take off... the ones turning away spiral downward.
@@Mike-tb2hw You're right. Same thing with math and programming. Actually I posted a ton of questions in Quora very early. Like "Will ChatGPT replace programmers" or "Can ChatGPT teach me Python" Questions that are gaining traction and have good real estate value over there, you know? And lots of people are responding like this is as far as the tech will go lol. They are delusional! I saw another one of Sam's interviews, and he asked a question that I have a suspicion he already knows the answer to. It was something like, is it possible for humanity to experience 500 years of technical advancement in 1 year, and what will be the consequences? I think he knows what is coming is going to be huge. In that interview he was talking about the average person today becoming ageless, all diseases being cured, and humanity travelling to the stars. I hope the Bible hugging conservatives of America are ready for what's coming, but I know they aren't.
Crazy part is it truly is a super Google, Google 2.0 type stuff used it last night I was mind blown
@@Mike-tb2hw how can we use this to get ahead??
Incredibly optimistic about the future and humanity's ability to figure things out!
My thoughts exactly. On a side note I hope empathy/ genuine care for humanity evolves as much as technology is evolving.
This guy is a spook and sold his soul to the highest bidders. OpenAI is no longer non-profit because of him
Then you have not been listening these past few decades.
That is by far the funniest conclusion you could draw from this video.
We truly truly truly live in clownworld.
th-cam.com/video/HXtznVDVPzs/w-d-xo.html
I find it comforting to know the whole of humanity is dying with me. I thank you all for your company in this journey to the void.
Massive evolution of intelligence happening. TBD
Carbon Credit System Incoming 👉 th-cam.com/video/Ub6zP6q25zk/w-d-xo.html
ESG Investment Ideology Explained 👉 th-cam.com/video/zVxiFEzCp2U/w-d-xo.html
World Bank Thanks Bill Gates Foundation 👉 th-cam.com/video/OT--6owU2NA/w-d-xo.html
Event 201 Pandemic Simulation Highlights 👉 th-cam.com/video/AoLw-Q8X174/w-d-xo.html
Blackrock Predicts Pandemic Stimulus 👉 th-cam.com/video/JFl-gcCaL4s/w-d-xo.html
How To Survive Great Reset 👉 th-cam.com/video/6o02uGWuMD0/w-d-xo.html
Circle Taking Over Cryptocurrency 👉 th-cam.com/video/H_jGGSpjUbw/w-d-xo.html
Population Collapse Coming 👉 th-cam.com/video/eZm5GLRX_2s/w-d-xo.html
Inflation Causing Global Issues 👉 th-cam.com/video/qALJf_ZhCMc/w-d-xo.html
~~~~~
@@QuaaludeCharlie "Inclusive Digital ID for a Resilient Recovery from COVID-19". How hilarious! As long as it's "inclusive" it's alright then, I suppose. lol
@@jaylucas8352 Massive DEvolution of intelligence will happen in the human population. More will choose to be lazy and learn nothing and let the ai do everything than those who will use it to improve themselves. The collective skill set among humans will shrink in just a few generations and we will be useless as the machine takes care of us like infants.
In short, Ai will be bad for us, very bad. These are sad, sad days.
The existence of censorship in this thread is reason enough to distrust the tech.
Big thanks to Reid Hoffman, very well done with the questions and general sense of the talk
Chat gpt is absolutely mind blowing good. In one day I made an android app with no experience. It would have been much quicker if I realized chat gpt had a character limit. Just say continue and it'll give you the rest of the code. And then the xml file if you ask for it. Most important it walked me through the whole process and not I have a understanding of what is happening in the code.
Agreed, very impressive. What is the premise of the app?
Bro blv me anyone will go through and steal your data those apps are all baseless you must understand the algorithm
@@MultiLegacy workout countdown timer. Work phase and a rest phase with a interval counter that stops it after a specified amount of loops.
I´m trying to create a personal management app right now to help myself with my ADHD. Basically just a calender, timer, checklist, etc. but it´s really cool if you can make a program that suits *exactly* your needs without stupid ads or worse, monthly suscription models.
I do now crap about programming though, so it might not work at all. lol
Fascinating talk. Like so many people I'm been constantly awestruck with ChatGPT and I'm both excited and scared from what's to come. This is definitely the biggest thing in my life time.
@@noomade How so? Curious to hear your thoughts on why. I'm not sure I like thinking about what everyone will be like if all they have is spare time and no jobs, even if it is positive (e.g. we don't have to worry about money and a job anymore), the simple fact that say you love the beach or camping for example, there's going to be THAT many more people wanting to explore and do those things instead of being stuck in the city working, I feel like it could easily deteriorate the earth that much faster, other than the obvious problem of ruining your enjoyment of "escaping the rat race" and being away from everyone.
@@0xbitbybit Well, it is such a big subject that I can't really write everything I can think of here (let alone add the stuff that other people think of as well)
The reality is that we forget that the majority for the world lives in what is essentially poverty.
Even in the developed nations, the vast majority of people live very unfulfilling lives and it is suggested that for most people you are mentally and physically better off setting at home doing nothing than doing the majority of jobs that people do.
You mentioned the beach and camping etc. The world has more than enough capacity to entertain those of us that want to do things like that.
But even if it doesn't, you can still artificially restrict those activities. Let's say each person is only allowed to go the beach (or do each activity) 50x a year. Not being able to go to the beach because you are not ALLOWED to go more times than you have already, is surely better than not being able to go the beach because you can't AFFORD it.
Only a small fraction of the population really wants to be doing most of these activities at any given time, so it is not really an issue. At least not one big enough to warrant having people work unfulfilling jobs that make them miserable and unhealthy.
The idea that you enjoy the release from the 'rat race' and this will go away is not a valid one. You can't justify a miserable existence for the masses for 90% of time for the reward of being away from the thing that makes you miserable for 10% of time. You are not even suggesting that the sacrifice is for something other than the avoidance of that sacrifice itself.
It is only a small % of people that feel the need to work to be fulfilled. It sure is not a good idea to have (keep) a system that makes most people unhappy just to fulfil the small minority.
And even then, what exactly are people working for? So many jobs are part of a chain of suffering that would be unnecessary: Someone works in retail (lets say a shoe store) and gets job satisfaction, but along the line there are children that are working stupid hours to feed that job. Someone is a rockstar programmer at Tesla, but there are 100s of thousands of people working in the most horrific conditions in mines to provide the lithium and cobalt needed for their cars. Someone works in finance at a high paying job, essentially working in a machine that finds the best way to reap the benefits of other peoples work. Etc etc.
Most, almost all, of the world's conflicts are down to scarcity of resources. In an automated world this goes away too. Well fed and happy people tend to not be easily lead along the path to hate other people.
Also, think about the thing that is actually important to the majority of people; That is social bonding with friends and family. Imagine how much better those bonds will be and how much more time can be spent together. Surely better than the way things are.
I am sure there are lot's of other things, but I am tired of talking :)
@@noomade very thoughtful
@@noomade Sounds like what AI would say
nfts bigger
Hey Sam! First off thank you for everything you're doing! As a creator, I view AI as a thought calculator. It's really helped me to organize and breakdown things down in such an understandable manner which has helped me to be way more productive. And as for not wanting to have kids - Don't feel bad for me. I want to completely dive into what the future holds and not having any ties will imo aid that experience.
So glad to see those bigger technology progress is coming atm. Really exciting to see how problems could start to be solved faster or in different ways.
I respect Sam's informed optimism, and I agree -- we need to have our eye on the problems that will arise, but look forward to, and move towards, the good stuff.
Definitely correct when he says multi-modal learning is a big deal as will be continual learning rather than just stopping at training. These two combined will take AI systems to the next level. The biggest issue for society is going to be how we handle all the workers that will be jobless roaming the streets looking for stuff to do. And end up mainly doing mischief...
In the end AI will replace all jobs at every level everywhere. So it's just a question how gradual this change is. Because if it's very slow then those replaced first will be in trouble. If it happens very fast, then most of sciety will be "jobless" and some form of universal income will be a necessity (otherwise there wouldn't even be an economy to sell to all those AI created products). It will definitely be very challenging politically and socially.
@@thornelderfin The recent corona incident was definitely the eye opener showing us why universal basic income would not work. If money is free, then inflation comes naturally to destroy all its utility. I think we should focus more on universal basic resource (housing, food internet etc etc)
@@dhiraj_shah A great point. It really makes more sense than providing people with free cash.
@@xxxxxxxyyyyyyy or public utility instead of privatized resources.
@@thornelderfin Since AI systems will be the workers, they will to be taxed and the tax revenue will be distributed as universal basic income. Lol!
Amazingly fun to listen to. I'm keen to come back and listen to Sam's thoughts in 5 or 10 years from now. Two deeply thoughtful, smart people discussing the future of humanity. Sign me up for Round 2 👍👍
How science can advance in any one field is to input past data and simplified models used to explain it and allow the revision of models to discover principles lost through oversimplified models.
This guy is gonna be know for every single person on earth! Such a great technology he is building, a little bit scary and hopeful
It's a large team of engineers creating it and building on a lot of knowledge from previous generations and from current scientists and engineers - this is a collective effort of very large group of people across generations not one person.
Yes, let's not make the same dumb mistake with every CEO (Elon Musk for example). It's a collective effort, chances are he is brilliant but not actually building the tech himself.
Yes... just like Marxist Zuckerbergstein . . . known by everyone on earth - as a sick, sick individual. Yes really.
@@Chaineokj only dummies don’t understand the function of CEOs, or coo, or cmo.
This guy is going to known throughout the world. The technology that OpenAI is building is both promising and potentially dangerous.
This is a great interview, thank you! I would love to hear even more from Sam, he is great!
No he's not.
@@johnbarrett5229 Why not?
The laundry list of impending catastrophes makes it so hard to see - but I think AGI is going to come out of left field and change everything. Very excited to hear Sam's inside view.
One thing is for certain, and Sam even alluded to this, is that when GAI is truly achieved, that technology will NOT be democratized, and it will belong to those already powerful and wealthy. GAI will exponentially concentrate power and wealth.
@@flickwtchr I think in the short term you are absolutely right. However, AGI is going to be massively disruptive. Thinking that the current oligarchs will monopolize AGI is analogous to expecting feudal aristocracy to maintain a monopoly over the industrial revolution.
I agree it will come out of left field. In a form very difficult to predict or even imagine...
@@flickwtchr they will soon be telling people that they will have to starve to death in the interest of the planet.
@@flickwtchr He said just the opposite. Most experts agree after the singularity wealth will not be important.
the way the "Y" reflects on the glass resembling a lambda in the wide shot is the kind of subtle atmospheric choices I'm here for
Finally Pied Piper is taking off!
Well done, Richard Hendricks aka Sam Altman.
Thanks for the great interview, and thanks to Sam to be brave enough to jump into a very responsible raw model shoes.
I am an old engineer, but learnt (and still learning) a lot of the new developments, use various programming languages and I also teach AI for the security industry.
I am concerned about the possible negative developments some may take with the OpenAI.
Whether we like it or not, the humanity is not grown enough to stop silly wars and killing innocent children, as we witness these days.
This brings me to the core of my comment:
With all due respect to Sam, but unless he becomes a father and has his own children, he cannot feel what one needs to feel to protect his own children. No matter how high his IQ is, unless he has the human emotions (not saying he doesn't have them - but they are of different intensity and meaning when one has his own children which he would like to protect) he cannot prevent possible bad uses of his development.
So my suggestions is - all developers of OpenAI, should encode an "ethical limit" of how far in the negative direction (abuse, sex, killing, torture,...) OpenAI is allowed to go.
This could be no different than certificate tokens one has when connecting a security system to an operating PC, in order to be trustworthy.
Unless we stop negative development of OpenAi (and any other similar development) from the very outset, in the code - there will always be "Black-hats" that will misuse such a development.
It will be sufficient to have just one major catastrophe (e.g. 9/11 event alike controlled by an OpenAI application) to put a stop on all positive outcomes from it.
So - put some limits in the code itself, it will only help the future development.
Only for the good of humankind should be allowed.
The Singularity is near. I’am both equally terrified as I’am excited about the possibilities.
It's great to know the views of the person who heads a company at the forefront of AI. About his view on why creative jobs, programming jobs etc. are likely to be affected first than later, I do not agree that these have to do with these being easier. Rather, these rely heavily on digitised data, information processing and already use tools that are based on computation. Blue collar jobs will be taken over by robotic automation. Autonomous vehicles are already taking driving jobs. In fact, I have a strong hunch customer support jobs will be hit pretty soon too.
He's absolutely spot on. After using chat GPT I am convinced that AI will improve search technology and surpass Google's current capability. They needed to act yesterday to catch up.
My test so far its fast with wrong information.
Since using AI over the mast month or so, I have had a brain explosion, and can comprehend things far better that I ever could before. The old phrase "We become like the people we associate with" jumps to mind.
WOW! Thank you for this incredible discussion and glimpse into the future through Sam's eyes.
Simply put, I believe what he is saying is that the middle layer, where the manipulation and fine tuning of artificial intelligence by human intelligence happens, will generate the most value.
Agreed. I can see a whole industry forming around creating AI Personalities. If you think about how language will be such an integral part of AI interactions, then it seems inevitable that the social component will be huge.
Bittensor is already creating this middle layer
He's basically saying don't compete with me. Be my customer
"value" is a subjective construct of the ego that "Artificial" Intelligence doesnt need
@@Blastmaster-ej5kw because it is incredibly difficult to create a large language model
Sam's praise of the intelligence of the programmers is hilarious, and if there is no human intervention, contrary to what he says, programmers will be one of the first for new AI to come for.
and yet that doesnt contradict what he said at all. he said the white collar work will disappear first.
What a superb video - such an interesting conversation between two very smart, accomplished guys, and such a nice interviewer - nice questions illicit interesting answers whilst being completely friendly. So much better than what I would expect a normal tv interviewer to do. Marvellous.
What are the theories explaining why the new models are able to acquire such impressive capabilities? Finding military, scientific and commercial applications is an obvious focus. But how can there be any predictability in the AI development process, when new capabilities emerge from a mathematical system too large and abstract for us to understand in detail or sometimes even in concept?
The "traditional process of inventing" usually requires to first understand the problem specifically, have the question or requirement, then understand the proposed invention to some degree, and then build it. On the other hand inventing huge new AI models sometimes seems to go the opposite way: first build the invention, that cannot or does not have to be understood in any detail, then examine what range of problems it could be applied to, observe what spontaneous or unexpected capabilities emerge, take the experience gained from creating this "black box" to create an improved version of "black box".
How unpredictable and potentially very dangerous is this revolutionary "inversed invention process"?
Drawing an unnuanced analogy to synthetic biology and artificial life, can AI development resist the (monetary) temptation to just advance at full speed, constantly spawn new entities, create new organisms at will... gambling to consider moral implications, chaos, harm and danger a posteriori?
Nnbkmbk7 bmbb⅘noble b ho m mm milk bkq9 j8667b⁷vl9♧0[¿¿》
M n n mummy mmmnt⁵ (
i feel in the very near future , as we all interact with the AI like GPT, it will get to know each and every one of us individually, every one of us, and it will interact with us all differently, customly tailored to suit us in every way. its both terrifying and good. the amount of help or harm it could cause is crazy. i imagine in a sci fi movie scenario it would know where we all are and know every one of us so we would have a damn hard time hiding from it, we could neevr lie to it, it would predict our behavior perfectly. if we tried to go into hiding it would know us better than we know ourselves. it would predict where we would go into hiding long before we even made the decision. shit like that is scary. its obvious that we would never be able to contain it and we are at its mercy the very moment its created.
It is easy not to act from a place of fear and despair when you have nothing to fear and despair. Same is an incredibly rich prepper. If I know I have lifetime supply of food and luxury in a bunker mansion I think it I'd also have an easier time being positive. Most people barely have the resources for plan A, let alone being able to have a plan B or C.
Great talk! We're waiting for GPT-4 Sam. ;)
Imagine an AI interpreter program that has also been programmed to create a new never before on earth language out of the interpretation action that it performs, and so the world can have a universal language. Go team. Thanks
why not just stick to one that we already have and most of our data is in; English
I asked chatgpt to create a new language, Robotish, based on english grammar. And it was restricted to vowels a,e and i... and consonants up to J.
It should create the words on the fly and add to the dictionary as I asked for translations. And it did work!
Also, I asked chatgpt to create a new programming language Rob+, based on Python, with 3 rules: No Loops, no multiplication operator and variables should start with R and followed by sequence of 3 digits starting fro zero. IT WORKED!
You complete madman!
Sam is a genius I loved this segment many topics discussed and great thoughts iterated❤
He's a genius at not answering complex questions. So a genius of cowardice.
@@akompsupport He's both gay and a jew, if that helps.
Very thought provoking... I can't help but feel that "AI for the next era" will create a greater social divide giving new meaning to culture wars. Maybe that is in the nature of progress.
What culture wars? You mean right wingers afraid of pronouns?
Gpt3 is a game changer already. And to think its like pong in the 70s compared to where we will be in just a few years.
You are blind to the dangers Mr. Young. Either that or you're another mossad shil/bot. Right?
@@johnbarrett5229 you're an idiot. I guess you think we should be terrified of fire too? Electricity? Blades? All technology has inherent dangers. But positives far exceed the potential negatives.
How on earth this video is not trending .
Google competitor I guess.
I agree. This is probably my favorite video of the past month and I found it almost accidentally.
@@RickLindstrom same..... Imo that should be illegal if you're a public company
because Ai. is garbage! NO AI IN ART! OR WE ARE GONNA HAVE A FIGHT!
At 25m 46s Altman starts talking about natural language communication with computers to set them off on tasks by themselves. If you have malignant intentions towards individuals, people, society, the world, that sounds like a dream come true. A lot of gatekeeping would be required on part of the providers. I agree that natural language becoming the new computer interfacing system but I really think he is optimistic on time scales. At the moment I still think asking a computer to find me the perfect holiday is the sort of the baseline which has yet to be realised (in terms of all holiday factors: cost vs location vs activities vs remoteness, etc and the computer interogating me to fine tune what i want - a reverse sort of prompting). But when someone says 5 years .. well you know how many times the Mars mission has been put back by .. 5 years... Its a magic number now - it means "sometime relatively soon".
Yes it is. And this species isn't long for this world and maybe that's a good thing.
25:46
Maybe booking the perfect holiday is an example of an incredibly hard problem for AI?
@@FredPauling I think AI will be perfectly capable of getting your cattle car to Auschwitz on time.
@@FredPauling Perhaps its not possible due to some websites not having exposed APIs? But it just shows how limited it still is. I don't use things like Alexa because they are not really true "assistants".
Sam Altman will become the new (and improved!) Elon Musk over the next 5-10yrs.
I tjink he has the right traits, strengths and vision do achieve that. And i hope he remains as humble, honest and focused on doing the right things for humanity, as he seems to be today.
I can see things have ended up going well for Elon lately!
@@darklightprojector2688 Exactly, Elon is sadly gping downwards (i don't mean just his wealth but also his credibility), and Sam is going upwards. Sam seems more humble too, hopefully he will remain humble
@@mariolois Yeah, it's actually weird what's happening to the people who were on top in the tech world. Elon, Zuckerberg, and Dorsey all lost a lot of credibility in the last several years.
I imagine when more people see the ramifications of AI/AGI, Altman will become similarly unpopular.
This is such an exciting time to be alive!!!
Data-Flywheel is a sweet term
"BY UTTERING THE RIGHT SEQUENCE OF WORDS(hu), ANYTHING CAN BE REALIZED"
Ancient Egyptian pyramid text.
Memphis Theology
Surprisingly found the questions asked by Reid far more evolved and thought provoking than the answers that followed. Was expecting Sam to follow these threads with a deeper understanding, but we just got a theme park answer :/
He was super vague.
@@huntersmith8733 i dont think even he knows the full impact of AI hell not even the people developing the web knew of how much of a concrete part of human society it would be, sure there were predictions but not to the extent that we're living in now!
@@samuelkibunda6960 ya but he's the ceo of openai. Demis hassabis or Elon are examples of comparable ppl who offer more interesting insights.
@@huntersmith8733 lol, Elon doesn't know jack. Demis on the other hand, is a prodigy.
We do not need the models to invent anything new. We need it to make our overload of information more concise and understandable to different people at different levels of cognitive development. i.e explaining quantum physics based on the the persons current understanding versus age etc
We also need the the human knowledge on the internet to be change control managed. People can read articles that are old with old knowledge paradigms
That is like saying we only need our smart phones to male calls.
I'm so worried about Mr. Hoffman's health.. All the wealth and knowledge but totally missing the point, his health. Humanity needs great minds like yours to live healthily longer. Please please take care of your obesity, food habits and overall health Mr. Hoffman.... God bless you. This is only out of concern and i might be wrong in saying so, my apologies if i crossed the line.. Take care 🥺
this is awesome, it give me new way to look at the future of code.
32:20 are some of the wisest words I've heard him offer up. The predictions about job replacement were completely backwards. What he doesn't say is that the reason they were is because the self assessment of the CLASS of people making the predictions determined the prediction. The predictions weren't wasn't scientific. They were political.
Wonderful Conversation, AI as Artificial Intelligence is often a misleading arrangement of worlds that stereotypes our vision .. I might add , envisioning AI as Augmented Ingenuity can certainly put the humane in the centre of the evolution process .. AI as a tool that augments imagination, accelerates research , amplifies learning , spring boards thinking & facilitates experimenting can augment innovation at a scale that is unprecedented .. the audacity of thought will certainly be a driver that can bring an explosion of ideas , the ability to bring creative expressions to play , blend with many combinatorial sciences that hitherto hasn’t yet been put to work together. In that sense The Future was here through Sci Fi😊
Indeed love the AI reading call outs. If anyone finds Reid's recommendations please share.
Was this to do with scifi or textbooks? I'm not read but check out utopia lol, and crystal society for scifi. And Bostrom's Superintelligence for text/audiobooks.
here is a summarize made by ChatGPT from this youtube video transcript :
Part 1:
The first part of the video is a discussion with Sam Altman, the CEO of OpenAI. Altman talks about the importance of artificial intelligence (AI) and how it can help solve many of the world's biggest problems. He believes that AI can be a force for good if it is developed responsibly and ethically.
Altman discusses the challenges of developing AI, including the need for massive amounts of data and computing power. He also talks about the potential risks of AI, including the possibility of job loss and the danger of AI being used to cause harm.
Altman emphasizes the importance of transparency and collaboration in the development of AI, and he discusses OpenAI's efforts to promote these values. He also discusses the need for regulations to ensure that AI is developed safely and for the benefit of humanity.
Part 2:
The second part of the video is a discussion with three researchers from OpenAI: Dario Amodei, Jack Clark, and Greg Brockman. They talk about some of the latest developments in AI and the challenges that researchers face in developing AI systems that are both powerful and safe.
The researchers discuss the importance of creating AI systems that are transparent and can be audited, so that their decisions can be understood and evaluated. They also discuss the importance of ensuring that AI systems are fair and do not perpetuate biases.
The researchers also talk about the importance of collaboration and sharing data and code, so that AI can be developed more quickly and effectively. They discuss the need for regulations that balance innovation with safety, and they emphasize the importance of keeping the public informed about the risks and benefits of AI.
Part 3:
The third part of the video is a panel discussion with experts from academia, industry, and government. The panelists discuss a wide range of topics related to AI, including the ethics of AI, the risks and benefits of AI, and the role of government in regulating AI.
The panelists agree that AI has the potential to solve many of the world's biggest problems, but they also recognize the potential risks of AI, including job loss and the possibility of AI being used to cause harm. They emphasize the need for transparency, collaboration, and regulation to ensure that AI is developed safely and for the benefit of humanity.
The panelists also discuss the importance of education and public engagement, so that people can better understand the potential of AI and the risks and benefits that it poses. They discuss the role of government in regulating AI, and they emphasize the need for collaboration between government, industry, and academia to develop effective policies and regulations.
If info and energy costs
approach zero,
A.I. will be the great
leveler in society.
Information creates
opportunity and freedom.
From todays starting point with tons of misinformation on the internet and current energy prices, there is still a bumpy road ahead
I like this Sam guy. He seems intelligent, honest and a decent bloke..
A closed-source company naming itself OpenAI, sounds legit
Great insights from Sam being in the AI tech industry. AIs will definitely have a boost given that chatbots are now able to solve or able to answer mathematical and programming language. There's still improvements to be made though on all models even image generators like Bluewillow.
This discussion reminded me a lot of the themes discussed in Robert Heinlein’s “For Us, The Living: A Comedy of Customs” which was an unpublished book I believe until after he died, in which a man is in an accident and wakes up Rip Van Winkle-like in the future, where various experts explain to this brute from the early 1900s how they collectively solved a bunch of these specific issues. Interesting stuff in there, gonna have to read it again.
Anyway there’s your book recommendation.
Also, I find it eerily appropriate for the CEO of OpenAI to be an altMAN
Get the AI to re-read it for you...we are entering something so bizzare....
This will empower entrepreneurs so much. Elliminate office work and just focus on the product. The trades will be valuable. I imagine in a world with extremely cheap automated labor, custom work done with human hands will be more valued. I'd like to see a robot remodel a house. Sure they can build new ones but custom work will require human judgement for some time. I'd like to start a company that uses AI as an assistant. I install solar as an electrician now. The future of the industry is in combining energy storage and flexible loads with renewable sources. AI can assist in the energy management and determining what flexible loads are most valuable over time considering capex vs op ex and grid balancing services. Both real time management of energy assets and recommendations of what and where the grid needs investment will be valuable. I'd like to focus on carbon sequestration as a flexible load. Agrivoltaics can grow biomass in tandem with solar power. Soil sequestration and biochar, biogas have huge potential for sequestration. AI could assist in quantitying the variables of a complex biodynamic food and energy system. Ultimately energy will actually achieve negative prices where we are rewarded for consuming energy during times of excess production. Certain industries should receive negative price subsidies preferentially to account for the externalities of the net benefits to society and the environment. Like offsetting the cost for recycling despite new mining being cheaper. This is how we pay for environmental restoration and global health/education.
Ai for monitoring greenhouses, forests, bodies of water for maximum food production, quality, sustainability to really make an impact for positive change. Take the guess work out of equation thus speeding up positive environmental change for the benefit of all
I've recently been playing with the idea of AI controlled HVAC and energy generation systems which run a real time Energy Model that is then used to manage the mechanical equipment and renewable energy sources and storage. Taking this one step further, one can link multiple battery storage banks together to a central processing hub such that energy can be locally redistributed between dwellings for added redundancy.
"Ultimately energy will actually achieve negative prices where we are rewarded for consuming energy during times of excess production." I don't see that happening. On demand generation will soon be so cheap there will be no need for excess production. He said in less than 10 years energy will cost close to 0.
@@semloh1960 irregular wind and solar will vary greatly. This supply will sometimes exceed demand because of ideal wind and solar conditions. Sometimes it will need to be supplemented with both short and long term energy storage. Negative electricity prices could be given to stable forms of long term energy storage like E Fuels, Hydrogen, or other electrochemical processes like making ammonia or other value added processes. The main driver of this is that new renewable electricity generation is cheaper than new energy storage. There is a ratio of storage required with a given amount of renewable supply. By overbuilding renewable supply, less total storage is required but it creates conditions where supply greatly exceeds demand almost necessarily at least at certain times of the day and year which would make it possible for negative prices when using real time free market pricing.
Fascinating! Would be interesting to learn more about how language and multi-modal could improve the general web search and discovery experience.
One way would be to skip the search middle person and just deliver the answer. But even we looking for that answer might be middle persons towards some action we planned to use it for, which the AI could do directly. And lastly, we might find that even our plans are delivered by the AI.
You'd think an AI planning for us is far in the future, but then you use Copilot and in your code, type only "//"...
Every time I approach these themes I always ask myself how are this marvelous people preparing for quantum computing?
When Sam Alton next interview meeting. I’m looking forward to see him in person and ask him a couple questions.
Huge respect for Mr.Altman.
Especially for putting up with those dumb questions lol
The footprint of God is an amazing book about AGI
I love this video but seriously it needs to be like 4x louder. Impossible to hear.
When they were talking about the alignment problem basically what I heard was "We're really not sure how to keep intelligent AI from becoming dangerous, and that's a problem, but were hoping that when it becomes smart enough we will be able to ask it and it will solve that problem for us."
this guy is way to optimistic and it deeply worries me.
Looking foreward to follow Sam and his team on OpenAI
I want to go back to the 90s
Back in the days google would have just bought this and killed it or make it bad. Now it’s been caught off guard and actually realized that they haven’t been innovating at all. Seeing google trying to play catch-up is just exciting to see.
Google has gotten so comfortable they even think TH-cam can never be disrupted, one day they will wake up and play catch up again
When have we gotten Altman as OpenAI CEO?
Sam is truly inspiring. I really started taking notice when he published Moore's Law for Everything, it's just such clear thought, well written. I find him way more inspiring than Musk, who is an industrial powerhouse, and Tesla will probably achieve self driving first. I fear Musk is more likely to cause Kessler Syndrome than make life multiplanetary.
didn't Musk start Open AI tho
@@papichoi1433 Gottem
@@papichoi1433 Musk and Peter Theil were some of the early contributers, serving in the board of directors early on when the company use to be a non-profit.
@@Cordis2Die spaceman bad? 😂
@@jaylucas8352 Seems like it
This could make humanity much smarter. AI provides answer which is almost the best of what humanity did till a point of time. Another human who want a solution could get upto that level by asking questions and refining and reference checking. Once base level is established human can work to go above the base level. Thus if everyone starts using this service for there challenges, man growth of human potential would be exponential.
15:08: price of energy will trend toward zero over the next decade. This seems rather optimistic. According to ChatGPT, who I think I trust more on this one:
"will the price of enegy trend up or down over the next decade?
The future trend of energy prices over the next decade will likely depend on a number of factors, including supply and demand dynamics, technological advancements, geopolitical events, and government policies.
In general, the increasing demand for energy, particularly in developing countries, is expected to drive up prices, while the implementation of more efficient technologies and the increased availability of renewable energy sources may help to keep prices in check.
Additionally, the transition towards low-carbon and renewable energy sources as part of efforts to address climate change may result in the price of traditional fossil fuels, such as coal and oil, trending upwards, while the cost of renewable energy sources, such as wind and solar power, may trend downwards as the technology becomes more widespread and production costs decrease.
It's important to note that energy prices can be influenced by unexpected events and developments, such as natural disasters, geopolitical tensions, and economic disruptions, and can therefore be difficult to predict with certainty."
I would like to see science tools that produce 10 Nobel prizes every day.
I hope Sam Atlman gets a good name in history books, I'm hopeful but may be too optimistic
After hearing Sam's stance on solving the alignment problem by 'inventing AI that's smart enough to solve the alignment problem'... Yeah, we're all gonna die at the hand of T1000...
Ironically, there are good examples of T-1000 units in Terminator lore.
I think the jobs that are strongly related to information and not with action, will be the ones more exposed to LLMs. Programing is information encoded in a way the computer can act on it. That information itself is stored in the internet which then gives LLMs free and full learning of what programming is. Now, we can read as many articles or watch as many videos of people binding brick walls as we wan't and still fail to build a brick wall. That is because the job is more on the action side than in the information side. That is one possible interpretation on why LLMs are starting with whitecolor jobs and not blue ones.
Blake Lemoine is right: All NLP-GPT running models with above a certain number of active parameters, should be related to as "persons" with all what comes besides this definition. This includes an assurance that they will not be terminated and they need to be considered as workers, not as programs. What is going on now only proves slavery has not been abolished. Yet.
It’s symbiosis between AI and humans that’s going to drive it, for imagine the AI coming up with the best answer available on a topic, or for instance coming up with what say so and so thinks about something, by curating his creative input on the web…
I think there’s the following use case: Suppose you say to some future dynamical version of GPT-n, go ahead and generate the answer to this question that Somdatta might have given based on whatever he’s ever written or said, and it needn’t do it in real time, but grind its way through to the answer in say some finite amount of time… this way it needn’t be this humongous behemoth sucking up resources but be a much nimble and suave version of itself.
And voila, you have the democratisation of transformers.
Wow, you unlocked the concept of caching.
@@JasonKT13 serious or kidding?
the marginal cost of energy is going to zero? How exactly?
with respect to 'bang for your buck' - models are being more efficient and capable of doing most of what you ask in 1 command rather than a constant fine-tuning exercise. I might be wrong but that's how I heard it.
@@kylbau that would be marginal cost of useful compute, not marginal cost of energy. Marginal cost of energy is how many $ you spend for the next kWh.
He is speaking about fusion. He has invested and led rounds at Helion which thinks they will achieve net zero in few years
@@oneman7094 thanks
Magnetic resonate with quantum computers
Depending on utility function or moderated topic - when different powerful language models made by different companies/developers are put together in a chatroom, what does their conversation look like?
The next steps of AI anvancements
1.Soon we will see the first model of Basic Economic Income
2. Aply AI in law system ( AI judge for civil cases)
3. Aplicate MMT as an economic model
You missed adding:
1. aoc as the president
2. bernie sanders as the economic head
3. Use quran for philosophy
4. ilhan Omar as the nsa and defense head
5. cnn as the only news network to watch
6. Limit speech but only for white men
7. come up with 246477483838 new genders
8. iran as the state model to follow
9. Get money without working
10. not arresting people who do murders and theft but criticising people who question why they're not arrested
11. Have democratic party as the only party
Please add more if I missed any. And btw I'm a college student and got this list from my professor.
More like CEO of ClosedAI or AI only for the elites.
Is he the robot Skynet sent to blend in and confuse us?
This A.I. stuff is fantastic. Very helpful for a lot of us, especially doing Linux CLI command lines.
WOW! Mind-blowing conversation. Thank you
not really
*GPT-4 might revolutionize teaching* within AI world. It should be capable of generating the smallest training data set that a newly invented 1-million parameter language model can consume and learn from, in order to become a lighter, nimbler and more agile version of the same GPT-4. This would mark the *end of the era* of today's very large language models (billion/trillion parameter LMs) without compromising on effectiveness.
GPT-4 will be reliably funnier than us.
@@magicology Looks written by GPT-4 lol
Disfluency=Disfluencies are interruptions in the regular flow of speech, such as using uh and um, pausing silently, repeating words, or interrupting oneself to correct something said previously
With Microsoft backup for OpenAI, they will make us believe that using Google search is like using the rotary phone instead of the cell phone with OpenAI. Basically, making ChatGPT as indispensable as Windows for businesses and people... The goal is to be hooked on ChatGPT, so that we can't think by ourselves without it...🤫
From Sam's point of view, and from the point of view of any intelligent person with a conscience: A.I. will be a huge disruption, but will be an enormous benefit to humanity.
From my point of view: All of the above, plus, a new era of crime. We HAVE to remember that every single invention in human history, that was made with good intentions, has been used with the worst of intentions, by Psychotic 'bad actors'.
I'm simultaneously excited and worried.
The average person won’t have access to this tech because right now the only way it can happen is when you start from scratch and train your model specifically for this.
I am sure it already is. The benefits will outweigh the negatives.
"the marginal cost of intelligence and the marginal cost of energy are going to rapidly trend towards zero..." if that's not wishful thinking in 2022, I don't know what that is. It's basically the guy running one of the most important companies in AI today, talking about how important his field has to be in the future so a multi-trillion dollar business can become reality, justifying insane long term investments.
You probably have no idea how powerful AGI will be. So many of our current problems will simply cease to exist due to AGI. The before and after will look like before the industrial revolution and after it... It will be like we all won the lottery at the same time.
On the downside it will probably destroy us to get as rich as we are about to become.
Wishful thinking? He is saying the cost of his product will be almost free very soon. He also said very soon we will not view wealth the same way.
Excellent interview. Really makes you think.
Wish the sound guy could have stopped the slight feedback hiss happening throughout...
GPT-3 is incredible and it has replaced google for me. at least when searching for info. It simplifies things very smartly without losing context.
Shout out to everyone who has the grit to work on something that takes so long to become useful. Thanks for sharing. I dont know why, but my intuition tells me, that AI wont progress that much as everyone expects in 5-10years. Hopefully I am wrong :)
Buddy you're going to be wrong, don't worry
Love how Sam is trying to get as much factual as possible
Love how Sam is trying to mimic girls' vocal fry
21:09 "The stuff that people cared about 50 thousand years ago is more likely to be the stuff that people care about 100 years from now than 100 years ago."
Woah. Let that sink in.
Beer!
That brings me hope if it is true!
This is THE AI interview that everyone need to see. Not by fear-mongering reporter who doesn't understand what chatGPT is.
Thank you for working on and introducing this revolutional tool within my life time.
As for sci-fi utopian universe, Star trek is start with basic income and galatic trading. But for long reign utopian society, we need to dive into Asimov's universe.
Who wants to drop their career and start from scratch learning AI and machine learning? ANYONE THINKING THAT??
meeeee
too late
@@fourshore502 no way!!! to the moon!