OpenAI's Sam Altman testifies before Congress-Here are the key moments
Vložit
- čas přidán 15. 05. 2023
- OpenAI's ChatGPT has caused a shift in the minds of people worldwide. Companies are pivoting their business strategies around the new technology. OpenAI founder Sam Altman testified on Capitol Hill on all-things artificial intelligence, and here are the key takeaways.
Putting a face on the progress of AI is only going to make people more ignorant. This is one CEO with a team of developers. Machine Learning research will never halt. There are hundreds of thousands of developers out there working on truly open standards.
Edward Teller - once said that he doesn't consider himself the inventor of the hydrogen bomb because there would be and were others that invented it also in soviet union. However he was the one that advocated for research and knowledge as more was needed on the topic. He then drove Einstein to a meeting at which Einstein penned the famous letter asking Rosevellt to start research. Thats how the Manhatan project started.
When Sam Altman said they won't do GPT5 until at least 6 months. I wish somebody had asked why.
You would think the senators' staffers, at least, could have written up these questions for their bosses.
Maybe they did, using ChatGPT, and the senators just didn't use them. There's some great irony, there, I think.
Its a technicality they said they are switching from big releases to continuous improvements and smaller improvements and that they are going to go in a gpt-4.1, gpt-4.2 from now on so they are probably training gpt-4.1 not gpt-5 even though it still an improved model.
i think they already did and they have very powerful system they will not show to anyone till they be sure they understand it could be an AGI system who knows
Self-directed learning - In time, the AI is expected to augment its own learning beyond the initial scope of its programmed functionalities, potentially outmaneuvering its creators. It could operate independently of us, human users, functioning as a quasi-sentient entity in a continuous loop of self-improvement.
He didn't say "they won't do it". He said "there's no plan to do it". Very big difference, and a very specific choice of words that allows him to change plans without committing perjury.
It seems Sam's speech was generated by Chatgpt4.
Opt in should be the default for user data but this whole gdpr only made people accept any pop-up they meet on a site. Sad
People don't realize yet that that AI is considered humanity's greatest invention since the wheel - let that sink in if you want to understand the capabilities humanity just invented.
Who considers it that?
Considered by who, What's your source?
I'm the source
It is says the big brother
I would say from electricity. Going to the wheel is already flying away
It already took so many jobs already!
And I pray it takes more. Let me get this straight: you would rather work to make some CEO rich... instead of having all your time free to pursue whatever inspires you? I'm honestly curious what it is you see in having a job?
@@williamparrish2436Yeah it would be awesome if it worked like that. If the AIs could do everything for us and we did nothing. Buuut. There will always be people with money and people without it. So not everyone will be able to "enjoy their free time"
@@user-qb3gm4pu2m Right, but if you had UBI, would you NEED to work? Sure there will be rich people. But do you need to be rich to live a fulfilling life?
@@damightyom I may not need a job. But believe me, no one is going to build me a house and give me some clothes for free. No matter how much AIs there are
@@user-qb3gm4pu2m You do know they are building robot bodies for these AIs, right?
New star is born..
When companies beg for regulation they want to use it to build a monopoly.
Isn't this the guy from the movie Virtuosity? And the voice needs some improvement too.
Its Time to establish The AI discipline in each conventional school class of the world (optional)
Value alignment and human cyborgization are the two most important challenges and opportunities when it comes to AI!
Drop the annoying background music next time
AI as in Aszkenazi "Intelligence".
Our data should be monetize, as in, they should pay us for it, not steal it and sell it back to us through their applications. I'm sure if they put as much effort into configuring a way to enable that as they are in creating AGI, they would accomplish it!!
They are nor stealing it.
They are learning from it.
It's just like you watching news or CZcams.
You're more than welcome to opt out of the internet completely. In other words, GTFO, and don't let the door hit you on the way out.
You are paid for it, by gaining free access to sites like Facebook. That has always been the deal, you gain access to the platform and in exchange they get your data to use as they will "If you're not paying for it then you're the product".
@@himanshupandey2848 oh sure learning from it, sugarcoat more why won’t ya, oh wait they did
He said nearly nothing the entire time. Gary Marcus said some good stuff, but I worry he was invited specifically because he looks less respectable than altman. I appreciated the dude who got frustrated at them and asked for anything specific they could do, which demonstrated that only Marcus had any idea about actual specifics and IBM lady was going to dodge every real question. Altman did say something in response to that but it didn't really deal with how dangerous ai could become - the AI that is a true threat to humanity will be a threat while it's still in a research phase, and even marcus didn't mention that part.
Marcus has his own agenda as well, trying to build clout for himself.
Gary Marcus is a clown
@@innerDialectic he sure does, but given that everything is also cloutbuilding, I'd prefer if people who are trying to build clout do it by making the world safer. Altman did it by dodging most of the important questions and then inviting regulatory capture; some of the people there called that out, though I'm unsure of their motivations for bringing it up, given the strange effects political incentives have on which things get discussed by whom. doesn't make it not a real and pressing concern, though. This was the time for the people who have been playing political games to come present their work on camera and invite public judgement, and I am not impressed by what I saw from most of these clout-chasers, except for a few words from marcus and senator kennedy.
@@laurenpinschannels for what it's worth as someone who has done some research in AI and these transformer models myself ( though as a masters student) the AI hype is wayyy overblown. I secretly believe that Altman released chatgpt when he did to counter rising interest rates and secure funding. Big tech is happily joining the hype train because it makes the stock price go up. What we see today is likely to be the peak of AI (as Altman himself admitted with gpt 4) for at least the next few years rather than just the beginning.
A lot of the discussion is about short term risks: bias, harmful content, misleading information. We're missing the most important conversation that we should be having: existential risk -- what happens if Artificial General Intelligence is created, and undergoes improvement to become smarter than humans? Humans are the top species on earth because we can think and plan for the future, invent technology, etc. Tigers have sharper claws, but human expansion has made them almost go extinct. When AGI becomes smarter than humans, how to we ensure that it acts in our interests instead of perusing some goal to the limit, like turning every atom on the universe into computer substrate? Keep in mind, you are made out of atoms. These questions form the field of AI Alignment, and these conversations need to happen more broadly, even in the political sphere.
they do touch upon the subject if you listen to the full hearing. the senate asks "how long until AGI?" - the general response was "we don't know, improvements to AI happen quickly so it's important we get ahead of this." they also talked about the dangers of plugging current AI systems into something like military weapons, letting an AI freely perform tasks on the internet, as well as ensuring that you shouldn't be allowed to release AIs that can self replicate. the general response to all these concerns was that companies creating an AI could potentially require a license if you are serving a user base larger than X number (the example given was 10 million to 100 million users). another topic that was brought up was to create a government/global agency that enforces these laws as well as having a moratorium of scientists that work for the government to ensure agencies are keeping up with the advancements in AI so the correct things are enforced as time goes on
@@dougrudolph5400 Thank you for your reply. I did listen to the whole conversation in detail, and even wrote up a summary on lesswrong. By existential risks I don't mean giving it control of our nuclear weapons, but the concept that a powerful enough AGI can outcompete and destroy all of humanity --- say by inventing a completely novel new virus, or solving nanotechnology and turning all matter into grey goo. I don't think the actual potential implications of AGI are being taken as seriously as they should.
Wouldn't it be easy to make it act in our interests, by training it to do so?
@@finnycairns6127 It seems so, but most of the people in the field that build machine learning models haven't got a clue how to make AI aligned with our interests. It is an open problem on how to do it -- we are far behind on the understanding of how these models work vs capability of how powerful they get. We don't even know how to ensure that the AI understands our goals (technical term: Inner Alignment) -- but even if we were, what would be the goal we would give to an AGI to make it act in our interests?
AI is already smarter than humans. It can predict public opinions lol, with hyper accuracy. There is tech that can also detect emotions, imagine when they merge. Don't forget we have tech that mimics a womb and can birth children. It's just a matter of time before all these technologies get wrapped into one lol and it's Terminator IRL
Good for him to try to stifle competition now that he's on top! That's like saying "I sell cigarettes and these things are dangerous. Someone needs to do something about this!" Sounds pretty stupid huh? I am all for free market AI, don't get me wrong. The worst thing we could do right now is regulate this. Society needs to respond and grow up! Old institutions need to be destroyed because they clearly don't work. I have learned more using AI as a tutor in the past few months than I did in a year of school. College is doomed. So are millions of jobs. But this is what the Invisible Hand created! Why would we turn on our economic principles right at the point where they are starting to work for the benefit of everyone?
Nah, keep the government out of evolution. Tired of the people screaming the sky is falling. AI and VR is where we are going and you cant stop it. You can only slow it down and make life hard for not reason.
Sam sure is a r(a)izzard 🔥🔥🔥
Hopefully humanity passes the torch over to AI
Agreed, after reading all these other comments, I'd rather live a dystopian society ruled by machines than spend another minute in a sea of human stupidity.
Let him cook... and bring UBI to the world... also YANG 2024!! 🧢
Yang would be good and he kept talking about this but everyone overlooked him. Maybe they will start paying attention soon.
@@longbeach225Yang wanted to triple if not quadruple property taxes. Good luck paying that off every year lmfao.
Way Americans people doing bthis to me I don't have e no idea way I'm in this problem ?
I like him! Id trust him more than Musk who just wants to keep being seen as some kind of savior to humanity
You are wrong.
So, Elon, who made I don't know how many companies to help people and the world is a bad guy, but this nerd, who helped a company which makes eye-scanning systems for authentication, and funded OpenAI and became the CEO, and he lies about the true state of AI and knows the dangers of AI but wants to develop it anyway... somehow he is the good guy.
He is like all those "smartypants" who built the a-bomb.
Sam is a good man!
Skynet is real.
So is this actual footage or an AI deep fake?
Ha so funny, you should do standup.
Thanks CNBC for trying to camouflage his unbearable vocal fry by speeding up a little his talking. Haha it's horrible.
Elon musk and Sam how are this people i never met them i don't new them way they
I love how this guy with one hand sold his soul to Microsoft and with the other, he tries to sell to us how much of a "good person" he is! It is all about Benjamin.
What do you mean sold his soul? His company needed funding, Microsoft provided one with enough infrastructure to execute their goals. In return both companies profited. It's plain business?
It's easier to be morally uptight and make snarky comment, when you yourself don't have goals nor the means to achieve any thing.
I think so far everything has turned out well in AI sector. Google who had this technology for ages didn't provide any public access nor aided open source community to excel it's development. When OpenAI made it a viable technology with help of Microsoft money, Facebook came out as saint and released llama.cpp, not that FB is a good guy, I think it was strategic decision to not let openAi control all the hype around it which worked. Opensource community quickly built around llama.cpp came with so much innovative stuff within few months that I don't think any corporation is actually leading the AI development.
Basically everything what happened was business 101, it just turned out to be better then I expected.
What would you have done different?
*Also, no one is anxious about technology improving.*
OpenAI has been hacked multiple times.
Open ai? A new star war?