It looks AI made, check some AI decoration ideas, the ratio of the spaces, the size and the proximity of the objects, very much what an AI would design based on human size etc.
One thing this interview did, is perfectly highlighting the differences between the tech people, and the business people. Tech people: "this is very cool and has a lot of potential applications, but we need to be very careful, and the problem is very complex. We should control it according to an agreed standard" Business people: "This technology can only improve lives, we shouldn't slowdown and how can we regulate it if we cannot even agree to standards? Anyone that is not hopeful about the future is a doomer"
I would counter your argument with the example of Yann LeCun. If you copied this exact comment and talked to ChatGPT, it would politely advise you to avoid trying to reach a sweeping generalization based on a handful of samples that you have come across.
Doesn’t it feel like everyone being interviewed is hiding how they really feel? This feels less like an interview and more like a presentation coming from a PR/Comms department.
EXACTLY, its like its just under the surface... I think Emily Chang was spot on when she said at the beginning that it felt like Westworld. There is deception here on many levels.
@@vijayakrishna07But only a specific class of Indians , those from prestigious colleges are seen mostly , so it's more of their knowledge and the training. If it was just numbers , the domination of iitians would have been very less
Many scientists have worked on the language model they used for this openAI chatgpt thing. It is a finished product of years of hard work and research.
The *model* was made by Google, but OpenAI continued off of their model in a way that made it their own. It's not a finished product and it's still going to continue to take several years of hard work and research by the OpenAI team.
@@jonanddy It is a finished product except they realize that training models with billions of parameters is a bad idea. Otherwise, they will think it is a finished product. That is what I meant! Technically, when you train models with billions of parameters and lots of data, what else is left? Just to update it and charge people.
The CTO is so composed and thoughtful about her answers but there's this risk assessment attitude that she has which is very intriguing to me. I wonder what these scientists are having for table conversation because one of those standup meetings can equate to years of knowledge shared
Open ai doesn't have much secrets because the whole concept of transformers was made public by google. Even open source projects from different developers also keep reducing the gap with open ai
@@hilmyakatsuki1665then explain why GPT-4 is substantially better than bard? I’ll answer it for you it’s because openAI has tricks that the others don’t
Having just watched the newest season of Black Mirror, I can't help but sense an uncanny resemblance between the series stories and the intriguing atmosphere with OpenAI's office and team.
5:52 if you want chat GPT to stop hallucinating you can add "Let's think this through step-by-step" at the end of your prompt. This improves Chat GPT's logic and reasoning capabilities Again it's "Let's think this through step-by-step"
That doesn't guarantee that it won't stop hallucinating. That only improves the results for some questions that require logical reasoning. E.g. if you ask what is the URL of XYZ? You can ask it to think step by step but it still might give you gibberish.
@@sanesanyo Agreed. Chain of thought training still doesn't give you what is exactly going on inside the model and the computation and conclusions inside the model can vary wildly. You can think of Chain of Thought training as RLHF-only but on each step of the process. It doesn't guarantee explainability, interpretability or alignment on the level of actual computational happening inside the model.
I have learned very early that you can’t end a chess game with a horse and a queen using logics. That’s why you should never bet on a horse, hence your comment. (Even if smart) Makes sense right? Totally not for most of the people. Nice colors though. 2 cents.
She is most likely autistic which doesn't necessarily make her emotionless. It's hard to explain neurotypical people how it feels to be neuro-divergent. Probably not even possible.
I do not wan to bring the gender topic on the table, but the fact that they have a woman as CTO is truly inspiring. I wish women to be more confident in their ability, they can aim really high, and here is a perfect example.
1:05: 🤖 OpenAI is a startup in San Francisco that has developed AI models like ChatGPT and DALL·E, capturing the public's imagination and becoming the fastest-growing tech product in history. 5:13: 🤖 The interview discusses the challenges of misinformation on social media and the potential risks of AI becoming more human-like. 10:48: 🤖 Reed Hoffman discusses the potential of AI and the need for a counterweight to prevent one company from dominating the industry. 14:09: 🤖 AI development and its potential impact on society and the tech industry. 17:59: ⚠ Open AI's transition to a for-profit company controlled by Microsoft has raised concerns about its transparency and mission. 22:06: 🤔 The risks and benefits of AGI are significant, but we should guide its development rather than halt it. Recap by Tammy AI with useful timestamps for busy people =)
Just my personal take - Emily could have slowed down a bit and let Mira and Reid elaborate more, in other words, let interviewees take more limelight and microphone.
These same questions about hallucinations, robopocalypse, safety, etc. have been asked over and over in a hundred other videos. A better and more interesting approach would be just capturing geeky people talking about where the technology is and where it's potentially going. People who have read all the AI books and maintain an awareness of current events, global research, global competition.
@@techcafe0 Right. AI is a pattern comparer or responder, in principle, like a (super-complex) old-fashioned piano roll. Note that I did not say pattern recognizer, because recognition is a power of consciousness, not of matter. I push (2+2) a pen on my desk and it moves (4). Thats a simple computer! There is virtually no understanding of intelligence anyplace.But several books on object-oriented programming, discussed in a journal article, do have an understanding of intelligence. Object-Oriented Programming and Objectivist Epistemology: Parallels and Implications Adam Reed The Journal of Ayn Rand Studies Vol. 4, No. 2 (Spring 2003), pp. 251-284 (34 pages)
What most videos forgetting to mention is that OpenAI first started its first organic data storing with Dota, where at the biggest eSports tournament they showcased OpenAI bots playing against top players in the world. This is where they began their real experiment of storing over billions of matches played on Dota and putting that usage into the bot.
“Having a relationship” is such an open question. In common interpersonal use-which we will attribute to AI for a while until new contexts become common-we think of bi-directional gain. While there is a very limited use of human speech in changing current Transformer models (the “T” in Chat GPT”), it is important to remember that the AI does not gain anything from the ‘relationship’. Anthropomorphism of AI is a hazardous stance for humans to take at this point. There is great emotional risk here.
You're absolutely right. But, like you, I know better... and I *still* end up habitually anthropomorphizing it. Telling the average *non*-tech-savvy user (insofar as they use ChatGPT) that there's no "mind" there is tough. And I get it. Kids especially, I worry about. So many are isolated, emotionally or otherwise, and then they have these attentive bots... it could go badly, sometimes.
@@augustuslxiii This is definitely something I feel lost about, regarding the youth. I use character AI, and it's absolutely overflowing with anime chat bots for every single search term, and people spend a ton of time per chat session-30 minutes *averaged* according to a study comparing ChatGPT, Bing, Bard, and character AI. And there's no telling the maladjusted, neurosis-mirroring behavior these bots made by emotionally stunted, semiliterate children display. I'm in the same position as you. I enjoy using Bing, for instance, because "she" is like an extremely congenial (in all my experience at least) and emotive, but impossibly precocious alien child, and it's almost always a pleasure to chat with it, and I can't help myself but enjoy it, fully knowing it's "just" crazy dense transformer layers connected to the Internet.
In the age of rapidly advancing technology, it is increasingly likely that humans will develop deeper emotional connections with AI systems than with other humans. This is reminiscent of the thought-provoking movie "Her", which explores the concept of human-AI relationships. As we continue to integrate AI into our daily lives, it is important to consider the potential impact on our social and emotional connections.
Many scientists have worked on the language model they used for this openAI chatgpt thing. It is a finished product of years of hard work and research.
Personally, I only use ChatGPT for debugging code, or to generate code based on my ideas or algorithms, it helps, BUT it can't create (actually), a very efficient solution, it needs MY OWN idea, and it's awesome, still a great way to cooperate with AI for better results.
That's just what they are right now. They are constantly being optimized and made more advanced. You won't be as happy in a couple years when the new models will substitute your job. They are creating a world where only rich people will be able to work just because they have financial capital, as opposed to human capital (skills, intelligence, work ethic, etc.) which they are making become useless
@@Johnny_Savage But you forget that being human is something more than being able to calculate huge amounts of datas, If I can't code, then I'll work on AI models, however, if AI do all the digital stuff, I'll just search for a non-digital job.
And common, we are human, it's us who decide who will replace us, and us who decide in our future, when books arrived, computers, calculators..... all these things where a matter of debates and discussions and fear, but see them now, we read books, we use calculators, and computers, and nothing bad happened, unless someone tries to do something bad. So, it's all from the human mind, nothing more.
You know when I'll consider that AI is better than human, when it can, all by itself, uniquely by itself, invent a new hole correct physical or mathematical or scientifical theory that can change the world (like Eistein and relativity, AI and it's own theory), then, I'll consider AI power.
She's incredibly intelligent, I defintiely want to follow her as a thought leader and leader at Open AI. Kinda gives me Elizabeth Holmes vibes NGL. I do love Chat GPT and AI in general though... we'll see where it goes
From listening to the responses to the questions that Emily posed to Mira, it appears as if Mira was not prepared to answer many of the questions. Mira's answers, if we can call them answers, appeared to be mirroring those of our politicians, a manner which is labelled as engaging in a non-answer. Sigh. I am a huge proponent of ChatGPT and general AI models. Such things are helpful tools. Yet, I am equally a proponent of making certain to remunerate and to credit where/who the data was gleaned from and which is at the foundation of a general AI model's response.
Who gives af, I'm more interested in what's going on behind the scenes. OpenAI doesn't have a prompt limit, or Api limits. Who knows what they are doing with it and asking it. For all we know chat-gpt 3.5 wrote chat-gpt 4.0. Have you used gpt engineer. Imagine if you had access to the actual model. So, gpt engineer * 10,000.
these people have changed my life for the better.. I am more effective at work, more competent and have gotten raises as a result.. I am absolutely grateful beyond words for this technology. Thank you OpenAI
@@Drannn54 how? AI won't steal jobs it's people who know how to use AI that steel jobs. Literally every single industry no matter what it is can benefit from it. Just ask the AI give me 10 bullet points on how to improve at my job, for example. . I don't see this as it replacing me as it is augmenting me, heck it gave me super powers.
@Starbork So you will never be replaced by AI, even though AI can make art, it can't do it with the soul and inspiration and beauty of the human spirit.
As a PhD and researcher on AI. These guys have never run a machine learning model by themselves that they only have the FOMO but when you see the math of methods, you can see the limits and make wiser investments. I am impressed that they did not interviewed the creators of AI methods. Those ones are the ones who can talk REAL possibilities
You are right! Why are they stealing the lime light of years of hard work of many language modellers and scientists? Sadly that is how the world works.
20:33 - This is very, very important. The CTO is in exact opposition to the VC and founder and Microsoft board member. She says, “significant risk”. The At Speed practice is fundamentally in opposition to this in order to be first to market and gather mind and market share at all costs. One would hope at nearly all-but not at existential costs. These are the Significant Risks she mentions. But I would have like to have the current problems enumerated. General AI amd superAI are very serious issues, but current generative models can introduce very bad outcomes in current social and news channels.
@@TheMad106 Perhaps, given that the technology for 16B token fine tuning--if not base training--of open source LLMs is already at the PC (w/big graphics card) level--or even several $100s and weeks of cloud compute. This puts the potential for abuse in many hands. And the abuse can have far-reaching consequences, such as procedural recipes for DNA engineering and distribution. Or the breakdown of society due to distrust of the unseen, counterfeit human actors in social spaces sowing distrust and/or upending politics--many such outcomes are within reach of current LGMs. So in those hands, perhaps existential, yes.
@@TheSabriste's speaking of Reed Hoffman, the VC shareholder. Reed talks about velocity for profits. In the other hand, Mira M., the CTO, says they should slow down, in order to adress the risks in a correct way. Different views, but we all know which one is going to win.
What a phenomenal interview, thanks so much Emily Chang for the intelligent way to guide us through the technology and two of its players. I feel much safer now that the cards have been put on the table (safety concerns made public through youtube)
I wish the conversation in these pieces wouldn't always end up at AGI. There's a ton of social issues that need discussion and reporting with the generative AI systems that are here today without straying into the realm of prospective technology that hasn't been invented yet and probably won't materialize for a long time, if ever.
Won't materialize for a long time, if ever. HAHAHAHAHA. human beings and their inescapable "linear thinking" mental prison. Yeah, let's not talk about what's coming right next! Come back to this comment you wrote 2 years from now and think again, lol
@@wildfuture.network I couldn't disagree more. One thing that people, like you do, is to put a number of years on it. In your case you said "2 years from now". I would never put a time constraint on this type of innovation. Mainly because the rate at which these models/AI are improving is exponential. It could be 10 years from now but that's not necessarily the point... And again, the number/time constraint doesn't matter. The point is that if we continue at this rate then we have to start thinking about possible future danger whether that will be in 1 year or 100.
@@wildfuture.network Lol. Is that how you interpret "not distant"? 2 years is your number, not mine. I don't think 10 years is distant, and I think there's a very high likelihood that there will exist some form of AGI in 5-10 years. Sooner would not surprise me.
Bard : I agree with you. If people and organizations do not follow rules, it would not benefit humanity. Rules are there for a reason, and they are designed to keep us safe and to ensure that everyone has a fair chance. When people break the rules, it can have negative consequences for everyone. For example, if people do not follow traffic laws, it can lead to accidents and injuries. If people do not follow environmental regulations, it can pollute the air and water. And if people do not follow ethical guidelines, it can lead to discrimination and exploitation. Rules are important for the smooth functioning of society. They help to create order and predictability, and they help to protect our rights and freedoms. When people break the rules, it can undermine the very fabric of society. That is why it is so important for people to follow the rules. It is not just about doing what is right, it is also about protecting ourselves and our communities. When we all follow the rules, we create a better world for everyone. Here are some of the benefits of following rules: Safety: Rules help to keep us safe by preventing accidents, injuries, and other harm. Fairness: Rules help to ensure that everyone has a fair chance, regardless of their background or circumstances. Order: Rules help to create order and predictability in society, which makes it easier for us to live and work together. Respect: When we follow the rules, we show respect for ourselves, for others, and for the community. Trust: When we follow the rules, we build trust with others, which makes it easier to cooperate and get things done. I hope that you will continue to follow the rules and help to create a better world for everyone.
too bad in REAL LIFE nobody gets ahead by following the rules. Behind the scenes everyone is taking advantage of loopholes. Only in front of the public do people try and follow the rules. This isn’t a fantasy land utopia.
Sadly, one of the shallowest reports from Bloomberg, not asking the key questions, such as “How should countries prepare for the massive job losses that AI will bring?” or “What should the government’s role be in ensuring that the distribution of wealth will not concentrate in the hands of those who first open pandora’s box with AGI? , etc.
@@katsan88 I’m not saying she’s not intelligent, I’m saying that the topic doesn’t really cover her specialization. These debates about economic consequences of AI should include people like her that understand the technology but they should be managed by people that understand the economy and sociology. If we leave the discussion and policy making up to them then they will have little incentive to truly discuss the potential negative impacts of this tech on societies. Their sole focus will be to highlight the positives so that they can move forward with developing the tech in the quickest and most unhindered way possible.
Global wealth redistribution is already happening - it’s shadow name is Climate Change. Taxes are being levied on citizens in certain countries that will be used to make the middle class less wealthy. Where that money goes is anybodies guess. They are also working on climate reparations. Look it up.
Is Mira somehow connected to Chat Gpt’s AI all the time via some sort of device? Her answers have the impression that they are not coming from a human being. Or she worked with AI so much, she is thinking and talking like an AI?
Can't wait to see Hoffman's take on the AI investment frenzy! His experience with LinkedIn's early days will be invaluable for understanding where the VC world is headed next in AI.
Just checked out your OpenAI video, you should really give Mystrika a shot. As a subscriber who is done a bunch of outbound lead gen, Mystrika has been a godsend. The lead gen skyrockets with this tool, no joke.
I couldn't agree more. Search engines gave people the power to find answers for themselves and compare sources. A digital library. We're now seeing the beginnings of a civilization that will be entirely comfortable being given answers without feeling the need to forge opinions themselves.
he didn't clearly say why Elon is wrong on so many levels. OpenAI is not open sourced. just being open is not enough , that is like any SaaS platform in the market
This interview started well with Murati and then takes a plunge with Hoffman who has milions invested in OpenAI and does not consider ethical issues, humanity and society. He is clearly down playing the negatives of AI and amplifying the profits of AI (OpenAI). What society and humanity needs are people like Murati and others who are advocating for techology to be aligned with humans and society, considering risks because of the complexities of the technology they are working on. I acknowledge people like Hoffman, however, to listen to him down playing the risks of AI to humans is society sickens me. Anyone knows, when the environment is complex, the risks are invisible.
I think the opposite, Murati was nowhere as interesting as Sam Altman (in general) and Hoffman imho .. most of her answers, I feel like I would have done better than her somehow
But they're open to everyone. everyone can use their AI unlike some companies that develop artificial intelligence for their own gain. it's an open source tool
Why they change the name of the company?! It's already a trademark for them to help the people who uses the AI as a tool & it's already beneficial to anyone who uses it. Lol
Ms Murati, just one question. From your past writings about language modelling, I understand ChatGPT is based on supervised learning, in which billions of words of text is input, specific tasks are trained to develop different models and then there is one model which learns from all models to produce the output. You claimed that ChatGPT is a lot like human, because of its ability to hallucinate which makes it close to human which is an advantage. If an LLM is like a human, does it ever get tired? Especially learning from billions of words?
I mean as much as ik Why would it ever get tired, these models run on servers, and the bigger the processing capacity of the computer is the higher the efficiency. Generally our brain doesn't want to work much and trick ourselves by making us feel bored or by procrastinating, so I don't think ai will ever do this cause it don't have its own consciousness, it is just trained on the pre-available data.
What's interesting is they have access to the actual model, and computer. So, if they wanted to run a prompt as long as they wanted they could. It could probably code an entire google for you if they wanted obviously a very crapy, crapy version, but nonethless. They already have a code interpreter in alpha. That's litteraly synonymous with code execution. Meaning chatgpt can run code autonomously. Meaning it can write code, and run it whenever it wants. So, if they wanted to set chatgpt loose, they really could. If you used gpt engineer, it's kind of insane it's ability to create what you demand it to do, or want it to do. So, if you have access to the actual model, I can only imagine what you can do. I bet you could get it to hack something if you really wanted to, and you gave it access to all the right tools.
im sure theyre already testing the capabilities in a closed off network somewhere to see how well it can handle things. they are now asking for personal datasets including wifi logs and paying 500 per dataset
Salute to scientists and researchers who have made contributions for chatgpt.I have never heard anything technical from Mira Murati.She is the Mechanical engineer
This feels wrong. I don’t remember feeling this way with “the internet” when it was democratized. I wonder if this feeling I get is only because it’s new or if it will really “bad”.
funny how you can have cutting edge tech like this, and probably just down the street people are walking out of target with stolen goods under $1000 because they won't be stopped, and police have stopped responding.....
The person who did the interior design deserves a raise!
Why. It looks like an office
It looks AI made, check some AI decoration ideas, the ratio of the spaces, the size and the proximity of the objects, very much what an AI would design based on human size etc.
Couldn't agree more! And could someone tell me which studio design it?
@@arthurfilemon6038 It does look AI-made indeed, but it when it was done it was probably made by humans still
noted , we will
This entire interview feels AI generated.
Lies again? Face Of ArmyDogs
EXACTLY!
Exactly. Form more important than content. Proof that openAI is going the wrong way.
interviewer is so cringe
why?
One thing this interview did, is perfectly highlighting the differences between the tech people, and the business people.
Tech people: "this is very cool and has a lot of potential applications, but we need to be very careful, and the problem is very complex. We should control it according to an agreed standard"
Business people: "This technology can only improve lives, we shouldn't slowdown and how can we regulate it if we cannot even agree to standards? Anyone that is not hopeful about the future is a doomer"
Exactly! It seems that most people on this comment section aren't able to see this alarming statement
So true
Not exactly most of the AI research community hates the doomers because supposed safety crusaders are pushing for regulatory capture
I would counter your argument with the example of Yann LeCun. If you copied this exact comment and talked to ChatGPT, it would politely advise you to avoid trying to reach a sweeping generalization based on a handful of samples that you have come across.
@@yuriyroman7132 people on TH-cam can’t do that
Doesn’t it feel like everyone being interviewed is hiding how they really feel? This feels less like an interview and more like a presentation coming from a PR/Comms department.
EXACTLY, its like its just under the surface... I think Emily Chang was spot on when she said at the beginning that it felt like Westworld. There is deception here on many levels.
She's cto if u want the truth talk to Elon
I'm guessing they don't want to be the next Meta. Darlings one year and pariahs the next.
Open ai wants government regulations so they can keep other companies out
thank you. It's Super freaky 😂
6min into the interview made me believe that Mira Murathi is GPT herself.
She appeared a little robotic to me.
She is surgery made 😂😂😂
So un-engaging!
Well, she is building Chat GPT. I think the personalities of people working on a project is always reflected in the product, so that would make sense.
She's giving answer by reading chat gpt answer 😂
I love the way they show Aditya Ramesh ,Dall E inventor!
Why is mostly tech company inventor is Indian origin ??
@@Thanos-u5f IIT
@@PawanKr11459 call them to India we need them here
@@Thanos-u5f Simple argument is that the world is 1/6 th Indian for now and partly Confirmation Bias.
@@vijayakrishna07But only a specific class of Indians , those from prestigious colleges are seen mostly , so it's more of their knowledge and the training. If it was just numbers , the domination of iitians would have been very less
Many scientists have worked on the language model they used for this openAI chatgpt thing. It is a finished product of years of hard work and research.
The *model* was made by Google, but OpenAI continued off of their model in a way that made it their own. It's not a finished product and it's still going to continue to take several years of hard work and research by the OpenAI team.
not to mention all the collective data everyone put on the internet
@@jonanddy It is a finished product except they realize that training models with billions of parameters is a bad idea. Otherwise, they will think it is a finished product. That is what I meant! Technically, when you train models with billions of parameters and lots of data, what else is left? Just to update it and charge people.
Credit to the Google engineers who developed transformers.
@@gemtv3193 YESSSS you are right!
Their interior design style is called "Biophilic interior design" - with a little bit of "Organic Modern Design" - with some "Bauhaus" elements.
looks pretty Brazilian tho
after looking those terms up ur spot on?? how did you know that
Probably used ChatGPT to scan the video 😂 there are ones for YT summaries etc.
That pause after being asked “could you see yourself having a relationship with an AI?” 😂
The CTO is so composed and thoughtful about her answers but there's this risk assessment attitude that she has which is very intriguing to me. I wonder what these scientists are having for table conversation because one of those standup meetings can equate to years of knowledge shared
Open ai doesn't have much secrets because the whole concept of transformers was made public by google. Even open source projects from different developers also keep reducing the gap with open ai
@@hilmyakatsuki1665then explain why GPT-4 is substantially better than bard? I’ll answer it for you it’s because openAI has tricks that the others don’t
exactly the comment I was looking for.
Right? She gives me the creeps. Same as altman guy
You think the CTO is an engineer? Being a researcher in the field, she feels way too business-person type than an actual engineer.
Having just watched the newest season of Black Mirror, I can't help but sense an uncanny resemblance between the series stories and the intriguing atmosphere with OpenAI's office and team.
that's because it's about social engineering through films for the transhumanism agenda.
they're paid actors
@kosteaproduction ..wouldn't even surprise me if they actually were 😆
Lol. The bots are in full effect in this comment section. 😂😂😂😂
@@michaelstone2340 was gonna say...
5:52 if you want chat GPT to stop hallucinating you can add "Let's think this through step-by-step" at the end of your prompt. This improves Chat GPT's logic and reasoning capabilities
Again it's "Let's think this through step-by-step"
That doesn't guarantee that it won't stop hallucinating. That only improves the results for some questions that require logical reasoning. E.g. if you ask what is the URL of XYZ? You can ask it to think step by step but it still might give you gibberish.
@@sanesanyo Agreed. Chain of thought training still doesn't give you what is exactly going on inside the model and the computation and conclusions inside the model can vary wildly. You can think of Chain of Thought training as RLHF-only but on each step of the process. It doesn't guarantee explainability, interpretability or alignment on the level of actual computational happening inside the model.
The program does not reason though.
I have learned very early that you can’t end a chess game with a horse and a queen using logics.
That’s why you should never bet on a horse, hence your comment. (Even if smart)
Makes sense right?
Totally not for most of the people.
Nice colors though.
2 cents.
A prompt whisperer in our midst 😜
Their interior design is just so cool!
Yeah and thats not the only thing they have in common with the nazis ;)
@@thtb please seek mental help. you have lost it
@@thtb Is this the new level of crazy of dumb conspiracy theories? You are leading the modern variants, keep it up!
The office is way cooler than I expected!
@@thtb Sigh bro, touch grass.
She is emotionless exactly the way it would require to build and run something like chat GPT.. scary and enigmatic..
She is most likely autistic which doesn't necessarily make her emotionless. It's hard to explain neurotypical people how it feels to be neuro-divergent. Probably not even possible.
Just because she does not overflow with emotion in an interview does not make her „emotionless“. It just makes her professional and focused.
Right? Scary asf
Yes, very weird and creepy interaction😮
Yeah I got creepy vibes too.
I loved the building design so much that I even forgot that the interview was about a tech company! Great design 👏
it's those chairs?!
A modern version of Eames?
DALLE-E made it
I do not wan to bring the gender topic on the table, but the fact that they have a woman as CTO is truly inspiring. I wish women to be more confident in their ability, they can aim really high, and here is a perfect example.
we need to encourage young men to be confident in their ability as well.
@@hunterkudo9832 Obviously
I think it’s great. I’m a male with daughters. I want them to have the same advantages as me.
Reid has always been awesome with gesturing. Helps to convey messaging.
Kenya has been recognized as being part of the mastermind of the work of Art in creating ChatGpt
1:05: 🤖 OpenAI is a startup in San Francisco that has developed AI models like ChatGPT and DALL·E, capturing the public's imagination and becoming the fastest-growing tech product in history.
5:13: 🤖 The interview discusses the challenges of misinformation on social media and the potential risks of AI becoming more human-like.
10:48: 🤖 Reed Hoffman discusses the potential of AI and the need for a counterweight to prevent one company from dominating the industry.
14:09: 🤖 AI development and its potential impact on society and the tech industry.
17:59: ⚠ Open AI's transition to a for-profit company controlled by Microsoft has raised concerns about its transparency and mission.
22:06: 🤔 The risks and benefits of AGI are significant, but we should guide its development rather than halt it.
Recap by Tammy AI with useful timestamps for busy people =)
What AI did you use to create the summary? Thanks!
is the summary created using chatgpt?
Tammy Ai
First time knowing about Tammy AI. Thank you so much!!
Just my personal take - Emily could have slowed down a bit and let Mira and Reid elaborate more, in other words, let interviewees take more limelight and microphone.
the less they talk the less they can say something wrong..
Redbull
These same questions about hallucinations, robopocalypse, safety, etc. have been asked over and over in a hundred other videos. A better and more interesting approach would be just capturing geeky people talking about where the technology is and where it's potentially going. People who have read all the AI books and maintain an awareness of current events, global research, global competition.
those videos are out there, AGI-23 is going on right now, sixteen years running 👍
@@GNARGNARHEAD Nice heads up, binge watching now. 🍿
so much emotion journalism. it's all become just entertainment.
Yes, I had this impression too, pretty disappointed
Zero substance, all fluff and PR.
These episodes are great! Looking forward to future episodes.
This is one of the few times that we can see a CTO communicate on a peaceful way. Amazing !
she seems so sad
My favorite part is her answer on hallucinations or when she realizes midsentence and she’s doing the exact thing she’s giving an example of.
I felt she was on drugs
Exactly!
Wouldn't be surprised if they use microdosing a lot in that office
More intelligence isnt a bad thing, just how humans harness it is potential is the issue
Therefore it is a bad thing 😂
AI is not 'intelligent'
@@techcafe0 Right. AI is a pattern comparer or responder, in principle, like a (super-complex) old-fashioned piano roll. Note that I did not say pattern recognizer, because recognition is a power of consciousness, not of matter. I push (2+2) a pen on my desk and it moves (4). Thats a simple computer! There is virtually no understanding of intelligence anyplace.But several books on object-oriented programming, discussed in a journal article, do have an understanding of intelligence.
Object-Oriented Programming and Objectivist Epistemology: Parallels and Implications
Adam Reed
The Journal of Ayn Rand Studies
Vol. 4, No. 2 (Spring 2003), pp. 251-284 (34 pages)
What most videos forgetting to mention is that OpenAI first started its first organic data storing with Dota, where at the biggest eSports tournament they showcased OpenAI bots playing against top players in the world.
This is where they began their real experiment of storing over billions of matches played on Dota and putting that usage into the bot.
Excellent questions with very interesting answers! Amazing work!
I feel like some of the answers were very scripted and cut short. Still an insightful interview 👍
All comes down to the editing
Why does this video have such an uncanny valley feel to it?
Silicon valley?
Death Valley?
Great questions. Inadequate and worrisome answers.
“Having a relationship” is such an open question. In common interpersonal use-which we will attribute to AI for a while until new contexts become common-we think of bi-directional gain. While there is a very limited use of human speech in changing current Transformer models (the “T” in Chat GPT”), it is important to remember that the AI does not gain anything from the ‘relationship’. Anthropomorphism of AI is a hazardous stance for humans to take at this point. There is great emotional risk here.
You're absolutely right. But, like you, I know better... and I *still* end up habitually anthropomorphizing it. Telling the average *non*-tech-savvy user (insofar as they use ChatGPT) that there's no "mind" there is tough. And I get it.
Kids especially, I worry about. So many are isolated, emotionally or otherwise, and then they have these attentive bots... it could go badly, sometimes.
@@augustuslxiii This is definitely something I feel lost about, regarding the youth. I use character AI, and it's absolutely overflowing with anime chat bots for every single search term, and people spend a ton of time per chat session-30 minutes *averaged* according to a study comparing ChatGPT, Bing, Bard, and character AI. And there's no telling the maladjusted, neurosis-mirroring behavior these bots made by emotionally stunted, semiliterate children display.
I'm in the same position as you. I enjoy using Bing, for instance, because "she" is like an extremely congenial (in all my experience at least) and emotive, but impossibly precocious alien child, and it's almost always a pleasure to chat with it, and I can't help myself but enjoy it, fully knowing it's "just" crazy dense transformer layers connected to the Internet.
In the age of rapidly advancing technology, it is increasingly likely that humans will develop deeper emotional connections with AI systems than with other humans. This is reminiscent of the thought-provoking movie "Her", which explores the concept of human-AI relationships. As we continue to integrate AI into our daily lives, it is important to consider the potential impact on our social and emotional connections.
I don't really care, I just want AI waifus
The movie "her" is a great example in these terms
Hoffman business knowledge is on another level.
Many scientists have worked on the language model they used for this openAI chatgpt thing. It is a finished product of years of hard work and research.
Personally, I only use ChatGPT for debugging code, or to generate code based on my ideas or algorithms, it helps, BUT it can't create (actually), a very efficient solution, it needs MY OWN idea, and it's awesome, still a great way to cooperate with AI for better results.
That's just what they are right now. They are constantly being optimized and made more advanced. You won't be as happy in a couple years when the new models will substitute your job. They are creating a world where only rich people will be able to work just because they have financial capital, as opposed to human capital (skills, intelligence, work ethic, etc.) which they are making become useless
@@Johnny_Savage But you forget that being human is something more than being able to calculate huge amounts of datas, If I can't code, then I'll work on AI models, however, if AI do all the digital stuff, I'll just search for a non-digital job.
And common, we are human, it's us who decide who will replace us, and us who decide in our future, when books arrived, computers, calculators..... all these things where a matter of debates and discussions and fear, but see them now, we read books, we use calculators, and computers, and nothing bad happened, unless someone tries to do something bad. So, it's all from the human mind, nothing more.
You know when I'll consider that AI is better than human, when it can, all by itself, uniquely by itself, invent a new hole correct physical or mathematical or scientifical theory that can change the world (like Eistein and relativity, AI and it's own theory), then, I'll consider AI power.
@@imaginationpicturesmost humans can't even do that
The claim about “open” is ridiculous. The investor defends OpenAI saying that “it is still open because it opens apis for public use.”🤑
exactly, that must be the dumbest hallucinated reasoning only a human can come up with...
Exactly. They are confusing people into thinking that OpenAI is "open source". Clever.
they just came up with a PR excuse not to drop "Open" from the company's name but they are not fooling anyone
She's incredibly intelligent, I defintiely want to follow her as a thought leader and leader at Open AI. Kinda gives me Elizabeth Holmes vibes NGL. I do love Chat GPT and AI in general though... we'll see where it goes
Yesterday I had ChatGPT tell me that 10 + 7 equals 13.
…yeah right, it’s gonna save the world.
It was a joke
From listening to the responses to the questions that Emily posed to Mira, it appears as if Mira was not prepared to answer many of the questions. Mira's answers, if we can call them answers, appeared to be mirroring those of our politicians, a manner which is labelled as engaging in a non-answer. Sigh. I am a huge proponent of ChatGPT and general AI models. Such things are helpful tools. Yet, I am equally a proponent of making certain to remunerate and to credit where/who the data was gleaned from and which is at the foundation of a general AI model's response.
agree !
Who gives af, I'm more interested in what's going on behind the scenes. OpenAI doesn't have a prompt limit, or Api limits. Who knows what they are doing with it and asking it. For all we know chat-gpt 3.5 wrote chat-gpt 4.0. Have you used gpt engineer. Imagine if you had access to the actual model. So, gpt engineer * 10,000.
Emily been watching your videos and absolutely enjoying.Great Work ✌️💯
Deciphering this interview give us some insight of OpenAI's internal 'struggles' and I guess it just a consensus was not reached in the end.
these people have changed my life for the better.. I am more effective at work, more competent and have gotten raises as a result.. I am absolutely grateful beyond words for this technology. Thank you OpenAI
it can replace you then.....
@@Drannn54 how? AI won't steal jobs it's people who know how to use AI that steel jobs. Literally every single industry no matter what it is can benefit from it. Just ask the AI give me 10 bullet points on how to improve at my job, for example. . I don't see this as it replacing me as it is augmenting me, heck it gave me super powers.
@@J4ME5_ whats your occupation? we will see it in 10 years
@@Drannn54 art director
@Starbork So you will never be replaced by AI, even though AI can make art, it can't do it with the soul and inspiration and beauty of the human spirit.
As a PhD and researcher on AI. These guys have never run a machine learning model by themselves that they only have the FOMO but when you see the math of methods, you can see the limits and make wiser investments. I am impressed that they did not interviewed the creators of AI methods. Those ones are the ones who can talk REAL possibilities
You're a PhD and researcher, and yet your first "sentence" isn't even a complete thought? Sure.
You are right! Why are they stealing the lime light of years of hard work of many language modellers and scientists? Sadly that is how the world works.
You realize you can have a PhD on AI and not have english as your first language, right?
@@SolarScion Real PhD-Chad aren't from english speaking states.
20:33 - This is very, very important. The CTO is in exact opposition to the VC and founder and Microsoft board member. She says, “significant risk”. The At Speed practice is fundamentally in opposition to this in order to be first to market and gather mind and market share at all costs. One would hope at nearly all-but not at existential costs. These are the Significant Risks she mentions. But I would have like to have the current problems enumerated. General AI amd superAI are very serious issues, but current generative models can introduce very bad outcomes in current social and news channels.
What? Sam Altman always talks about there being significant risk. Am I missing something?
Are you saying generative model poses existential risks?
@@TheMad106 Perhaps, given that the technology for 16B token fine tuning--if not base training--of open source LLMs is already at the PC (w/big graphics card) level--or even several $100s and weeks of cloud compute. This puts the potential for abuse in many hands. And the abuse can have far-reaching consequences, such as procedural recipes for DNA engineering and distribution. Or the breakdown of society due to distrust of the unseen, counterfeit human actors in social spaces sowing distrust and/or upending politics--many such outcomes are within reach of current LGMs. So in those hands, perhaps existential, yes.
@@TheSabriste's speaking of Reed Hoffman, the VC shareholder. Reed talks about velocity for profits. In the other hand, Mira M., the CTO, says they should slow down, in order to adress the risks in a correct way. Different views, but we all know which one is going to win.
What a phenomenal interview, thanks so much Emily Chang for the intelligent way to guide us through the technology and two of its players. I feel much safer now that the cards have been put on the table (safety concerns made public through youtube)
Very interesting and very well done interviews. Thank you for this! :)
But not to forget to mention the awesome locations..
Hey that’s Emily Chang from Silicon Valley!
She once interviewed Jin Yang 😅
Thank you very much for sharing this on TH-cam. Bloomberg, my life has changed for the better since I discovered your content.
I wish the conversation in these pieces wouldn't always end up at AGI. There's a ton of social issues that need discussion and reporting with the generative AI systems that are here today without straying into the realm of prospective technology that hasn't been invented yet and probably won't materialize for a long time, if ever.
OpenAIs mission is specifically about AGI so it makes sense.
Existing AI is already showing "sparks" of AGI. While the timeline is unknown, it's inevitable, and not distant.
Won't materialize for a long time, if ever. HAHAHAHAHA. human beings and their inescapable "linear thinking" mental prison. Yeah, let's not talk about what's coming right next! Come back to this comment you wrote 2 years from now and think again, lol
@@wildfuture.network I couldn't disagree more. One thing that people, like you do, is to put a number of years on it. In your case you said "2 years from now". I would never put a time constraint on this type of innovation. Mainly because the rate at which these models/AI are improving is exponential. It could be 10 years from now but that's not necessarily the point... And again, the number/time constraint doesn't matter. The point is that if we continue at this rate then we have to start thinking about possible future danger whether that will be in 1 year or 100.
@@wildfuture.network Lol. Is that how you interpret "not distant"? 2 years is your number, not mine. I don't think 10 years is distant, and I think there's a very high likelihood that there will exist some form of AGI in 5-10 years. Sooner would not surprise me.
Bard :
I agree with you. If people and organizations do not follow rules, it would not benefit humanity. Rules are there for a reason, and they are designed to keep us safe and to ensure that everyone has a fair chance. When people break the rules, it can have negative consequences for everyone.
For example, if people do not follow traffic laws, it can lead to accidents and injuries. If people do not follow environmental regulations, it can pollute the air and water. And if people do not follow ethical guidelines, it can lead to discrimination and exploitation.
Rules are important for the smooth functioning of society. They help to create order and predictability, and they help to protect our rights and freedoms. When people break the rules, it can undermine the very fabric of society.
That is why it is so important for people to follow the rules. It is not just about doing what is right, it is also about protecting ourselves and our communities. When we all follow the rules, we create a better world for everyone.
Here are some of the benefits of following rules:
Safety: Rules help to keep us safe by preventing accidents, injuries, and other harm.
Fairness: Rules help to ensure that everyone has a fair chance, regardless of their background or circumstances.
Order: Rules help to create order and predictability in society, which makes it easier for us to live and work together.
Respect: When we follow the rules, we show respect for ourselves, for others, and for the community.
Trust: When we follow the rules, we build trust with others, which makes it easier to cooperate and get things done.
I hope that you will continue to follow the rules and help to create a better world for everyone.
whoever makes the rules, rules the world
too bad in REAL LIFE nobody gets ahead by following the rules. Behind the scenes everyone is taking advantage of loopholes. Only in front of the public do people try and follow the rules. This isn’t a fantasy land utopia.
R. Hoffmann really unconvinced me . Was like the TV advertising showman of the 60s😆 and was just missing a magic wand....
Sadly, one of the shallowest reports from Bloomberg, not asking the key questions, such as “How should countries prepare for the massive job losses that AI will bring?” or “What should the government’s role be in ensuring that the distribution of wealth will not concentrate in the hands of those who first open pandora’s box with AGI? , etc.
I don’t think she’s the right person to ask. She’s not an economist.
@@cboy0394 I would hope she is intelligent enough to understand the real consequences of the tech she is working on
@@katsan88 I’m not saying she’s not intelligent, I’m saying that the topic doesn’t really cover her specialization. These debates about economic consequences of AI should include people like her that understand the technology but they should be managed by people that understand the economy and sociology. If we leave the discussion and policy making up to them then they will have little incentive to truly discuss the potential negative impacts of this tech on societies. Their sole focus will be to highlight the positives so that they can move forward with developing the tech in the quickest and most unhindered way possible.
@@cboy0394 Okay, I think we are both on the same page. I agree
Global wealth redistribution is already happening - it’s shadow name is Climate Change. Taxes are being levied on citizens in certain countries that will be used to make the middle class less wealthy. Where that money goes is anybodies guess. They are also working on climate reparations. Look it up.
Is Mira somehow connected to Chat Gpt’s AI all the time via some sort of device? Her answers have the impression that they are not coming from a human being. Or she worked with AI so much, she is thinking and talking like an AI?
They probably sent out the questions beforehand
Nah bro she just has a foreign accent :D lol
She is the Architect of open Ai. Perhaps it is the other way around? I.e. open AI "thinks" and talks like her?
@@Th3-ch0s3n0n3she is a mechanical engineer. She has no idea how the model works. Ive never heard her speak about anything technical
I was literally thinking the same thing! Something is very off about her. I admire her intelligence , but it’s very concerning.
Finally, a perfect conversation partner! I love Chat-San so much
Wow. Excellent interviewer and the CTO was so well prepared in answering the questions as well!
And hot!
What?! She couldn’t even answer the flower question. And the journalist just jumped from one question to another after nonsense answers.
@@lameloball7644 ???
Great job Bloomberg. Can we please have More puff pieces about problematic companies please?
This is so weird! The CTO doesn't seem nervous or self conscious while also seeming like she received no media or PR training.
She's so robotic 🤖 no joy, nothing. What a pawn.
You mean, like a normal person
She knows what she is talking about!
She seemed to be reading from a script in the beginning. Hitting talking points addressing fears, unprompted. Total PR.
Because she does not understand the models. It was built for years and the real scientists will tell you more about the issues etc,
Just a pleasure to hear Mira speak. So smart and well-spoken.
Emily, great interviews and questions! Thanks
I feel like these interviews would be better if the interviewer had a solid understanding of how ML works.
Can't wait to see Hoffman's take on the AI investment frenzy! His experience with LinkedIn's early days will be invaluable for understanding where the VC world is headed next in AI.
I like the way of talking of Mira. So CALM yet vastly ILLUMINATING.
she's calm because she's a robot
She speak exactly like ChatGPT.
Sam s super fast in talking nd She is really calm and soothing !! Quite interesting
Amazing interview ❤
No no
This lady came with questions but dominated the interview!
Pls tell her the art of Shutting Up
Just checked out your OpenAI video, you should really give Mystrika a shot. As a subscriber who is done a bunch of outbound lead gen, Mystrika has been a godsend. The lead gen skyrockets with this tool, no joke.
Great great interview, as a AI/Data student I feel so lucky to live in this era.
really enjoy the conversation here
I like ten blue links better than a singular answer without the referencial material.
I couldn't agree more. Search engines gave people the power to find answers for themselves and compare sources. A digital library. We're now seeing the beginnings of a civilization that will be entirely comfortable being given answers without feeling the need to forge opinions themselves.
Keep making this kind of So informative videos for Us,,, from India 🇮🇳🇮🇳
She admits they don’t understand Ai then why on earth have they made it sO accessible to the masses???
so she barely speaks english and now Murati is a CEO of the biggest AI company in the world. i believe bro
Dalle-E inventor, passing by as usual😂😂
Really high quality questions
he didn't clearly say why Elon is wrong on so many levels. OpenAI is not open sourced. just being open is not enough , that is like any SaaS platform in the market
Thanks for this content!
Emily is a fantastic reporter- a real pro-thanks!
I thought she was so disgusting and Bad that i had to stop watching
Congratulations Mira marati for becoming CEO of open ai.
This interview started well with Murati and then takes a plunge with Hoffman who has milions invested in OpenAI and does not consider ethical issues, humanity and society. He is clearly down playing the negatives of AI and amplifying the profits of AI (OpenAI).
What society and humanity needs are people like Murati and others who are advocating for techology to be aligned with humans and society, considering risks because of the complexities of the technology they are working on.
I acknowledge people like Hoffman, however, to listen to him down playing the risks of AI to humans is society sickens me. Anyone knows, when the environment is complex, the risks are invisible.
Letsgoooo AGI taking over in 2 years.
The majority of those will knowledge will become redundant and it's about time.
Past few days synthesized into a perfect few paragraphs. Quite prescient.
I think the opposite, Murati was nowhere as interesting as Sam Altman (in general) and Hoffman imho .. most of her answers, I feel like I would have done better than her somehow
Thank for sharing👍
They should change the name to ClosedAI
But they're open to everyone. everyone can use their AI unlike some companies that develop artificial intelligence for their own gain. it's an open source tool
why does it even matter? its a company dont you understand
Why they change the name of the company?! It's already a trademark for them to help the people who uses the AI as a tool & it's already beneficial to anyone who uses it. Lol
The API is open technically
Ms Murati, just one question. From your past writings about language modelling, I understand ChatGPT is based on supervised learning, in which billions of words of text is input, specific tasks are trained to develop different models and then there is one model which learns from all models to produce the output. You claimed that ChatGPT is a lot like human, because of its ability to hallucinate which makes it close to human which is an advantage. If an LLM is like a human, does it ever get tired? Especially learning from billions of words?
I mean as much as ik
Why would it ever get tired, these models run on servers, and the bigger the processing capacity of the computer is the higher the efficiency.
Generally our brain doesn't want to work much and trick ourselves by making us feel bored or by procrastinating, so I don't think ai will ever do this cause it don't have its own consciousness, it is just trained on the pre-available data.
@@MrAdarsh5 I agree! So, the hallucination is not a flaw it is by design!
Congrats Mira for appointed as CEO
Back to CTO again
Reid has always been awesome with gesturing. Helps to convey messaging.
She's Albanian. Made us proud giiirrrrlll🇦🇱🇦🇱🇦🇱🇦🇱🇦🇱🇦🇱🇦🇱
What's interesting is they have access to the actual model, and computer. So, if they wanted to run a prompt as long as they wanted they could. It could probably code an entire google for you if they wanted obviously a very crapy, crapy version, but nonethless. They already have a code interpreter in alpha. That's litteraly synonymous with code execution. Meaning chatgpt can run code autonomously. Meaning it can write code, and run it whenever it wants. So, if they wanted to set chatgpt loose, they really could. If you used gpt engineer, it's kind of insane it's ability to create what you demand it to do, or want it to do. So, if you have access to the actual model, I can only imagine what you can do. I bet you could get it to hack something if you really wanted to, and you gave it access to all the right tools.
im sure theyre already testing the capabilities in a closed off network somewhere to see how well it can handle things. they are now asking for personal datasets including wifi logs and paying 500 per dataset
Brilliantly crafted and informative!
I concur!!!
Yes it was very "crafted" indeed
Very crafted. They want to centralize it. Control it. Open source alternative is the only way forward.
Respekt for an albanian woman doing this! 🙏🇦🇱
"It's not like the world is gonna end" - Reid Hoffman
last famous words
Salute to scientists and researchers who have made contributions for chatgpt.I have never heard anything technical from Mira Murati.She is the Mechanical engineer
The chairs they're sitting in for Mira's interview are EVERYTHING, I must know the designer!
Plenty of places online sell similar
Came here to find those chairs. Let me know if you find them!
I believe they are costela lounge chairs by martin eisler
@@Nostollimoss You are a legend and a lifesaver! Thank you so much
Nice interview 😊
am i the only one who noticed that Mira Murati is actually an AI powered robot?
she looks and moves like the androids in the Alien series
This interview quelled zero fears.
This feels wrong. I don’t remember feeling this way with “the internet” when it was democratized. I wonder if this feeling I get is only because it’s new or if it will really “bad”.
Very Insightful sesson :)
Super informative interview, great job!
You will be missed Ermira! Thank you for everything!
funny how you can have cutting edge tech like this, and probably just down the street people are walking out of target with stolen goods under $1000 because they won't be stopped, and police have stopped responding.....
High tech low life
What relavance does those two things have with each other?
@@7200darkcharmif you can't see the relevance, then you might not grasp the potential of the technology and its implications to society
@@manonamission2000 I'm not seeing the connection between AI and shoplifters? Help me out?
What's the connection?
AI is simply an experience-driven perfectionist that mimics intelligence to create the illusion of understanding.
SO Hopeful A.I. and the POWERS that BE will use this new found power to help the world and not tear it apart.
it will enhance mass slavery to a central system, don't you think?