If all engineers walked away from maintaining a large model like GPT-4, the model itself might continue to function in the short term. However, over time, issues would likely arise. These could include server outages, data inconsistencies, or security vulnerabilities, eventually causing the model to become ineffective or even cease functioning. So, while it may run autonomously for a while, long-term operation requires ongoing maintenance and oversight.
The film "2001: A Space Odyssey" already warned us about a malfunctioning AI named HAL-9000 where the human operators didn't listen to its warnings. HAL warned the astronauts that it was acting "silly" and unsure of itself even though in the interview it advertised itself as "fool proof" with no prior errors.
I see a serious need to use the same degree of "safety" needed for AI to be used in picking future leaders who will be debating policy. Another question I have is what is the ultimate goal of developing this high level of AI once we remove selfish greed and gratification from the list of our goals? From what source of ethics are we going to use to make AI policies from? I have so many questions that I would love to hear answers to. Thank you for this awesome interview.
To simply put it - Machine learning models learn from data and vary in complexity and cost. Large models like GPT-4 are expensive and need ongoing maintenance.👨🔧👩🔧
There’s a better video example “openAI plays hide and seek and breaks the game” That’s a simple explanation for the future of war strategies. I have more if ur interested.
Theory:Hamas used AI designed for war by the regime in Tehran to create the most optimal strategy for the assault.Just like old AIs finding creative solutions to games now the new ones find creative solutions for wars .AI wont destroy the world,we will by asking them how!
Guys… machine learning models are currently useless without human input. Yes, In the wrong hands these models could be very dangerous and that should be looked at carefully. It’s important to prevent nefarious actors from developing their own models.
@@northsky5629 I believe the near future of machine learning models in the USA 🇺🇸 is going to be, to consolidate waste data, taper the cost of software dev, continue public market research, innovations in established ML systems, innovations in production and manufacturing, and probably a ton of other statistical analysis. It’s possible that the Google employee left do to multiple reason.
AI scientist at a big 5 here. The people that sign these "letters" and do all these public engagements have not written a line a code in decades. This stuff makes 0 sense
Tell China and Russia to slow down before we do. It’s similar to any high tech military weapon. Every teacher and kid needs to be educated, no excuses for being old.
Im a bit mixed on it, i feel like AI is rapidly developing in certain fields, especially artistic fields, but i also get hard "nuclear fusion is only 10 years away!" vibes.
I use ChatGPT 4 almost everyday to do complex data analysis daily by providing it screenshots of performance and ask it for analytics as well as focus areas and solutions. From my experience, it is moving at light speed. The difference I’ve seen in less than one year is astonishing.
Lookup Alex Karp. Lookup his public schooling history. How many languages? Lookup the Organizations, and Boards he has been member of before 30yr of age. You know Palantir, right? You know In-Q-Tel? You know Lockheed/FILecoin? An M-4 will mean nothing by 2025.
So we take a respite on breakthrough development on AI but North Korea advances and does terrible harm to the world. Or, do we make the breakthroughs that could defend against whatever threats pop up?
I was thinking about this all militaries will have to go the way of agi designing and conducting war. If not the military would be behind other nations. All nations will be doing this.
I think people are far too worried. Something could go wrong with nearly anything. How does a calculator know how to do what it does? Because a human being put the information onto a thing and put it into the calculator and then, the calculator does what it does and fast. This AI did not make itself, it was made the same way the calculator was made, that would be by a human being or another machine that a human being made and AI will only know whatever information a human being put into it for it to know and that will be all it will ever know and all it will be able to do is what a human being made it to be able to do unless somehow something evil can also do that, which i doubt, but it knows what a human being put into it to think about when someone asks a certain question and this and that and it could jumble things around and that's all it can do therefore the biggest threat to this planet is, ever since Adam and Eve, human beings and their sin mixed with influences from Satan and his other fallen followers and people like to pretend that life is whatever they feel is life in their head and make up reasons for everything that they feel like is right inside of their little brain and I mean this with all due respect and I am fully aware that I am also a human being LOL I say little brain in order to give a perspective on just how much a human being does not know compared to God, which would be Jesus Christ as described in the 1611 King James Bible, which is actual recorded history as well as future and you can go see that for yourself with your own two eyes. Put your faith on what Jesus Christ did for you and no matter what happens in this life, you're going to be just fine or you can keep on pretending that whatever you feel is right is right and be wrong all the rest of your life until you die and go to Hell and that choice is completely up to you and you'll do what you want to do either way, but as for me, I have to have proof and evidence for these vastly important things before I'm going to be believing them and I will believe anything that can be proven and thus far at 36 years old about to be 37 and of sound mind, I fully believe the King James Bible because of the proof and evidence that goes along with it and the deafening silence instead of proof and evidence that comes from everything else trying to tell me where human beings and everything around them came from. Things could go wrong with something a human makes and depending on how much power a human gives this new thing, that a human or multiple humans made, it may cause a big problem or a little problems or just be talked about, but it's definitely not going to end this world and I know this because everything the King James Bible said was going to happen has happened and it does not tell me about a future where AI destroys the planet or all life on the planet therefore I know it's not going to. God aka Jesus Christ is going to destroy this Earth Himself and make a new one after that, but many different horrible things are going to happen before that time comes, maybe AI has something to do with some of it and maybe not, but some will experience them and some won't, but if you actually want to know what's going to happen and you actually want answers to your questions about what this life is for and where everything come from and such and not just some false answers that you want to pretend are true because you really want them to be, go see what the King James Bible says and you'll know. The Gospel (Good News) 1st Corinthians 15:1-4 KJV faith fully on that is humanity's only hope.
To come up with something like AI in the first place may seem like a very intelligent thing to do, however, it turns out that it’s the pinnacle of stupidity.
On a deeper level everything that can happen does. Therefore in other universes it is happening now. The only question you need to ask is are there lunatics running the show dumb enough?
I'd rather listen to someone who is an actual AI engineer, not a policy Lawyer. When speaking about AI becoming sentient you must ask yourself one question. How can we create something that is sentient and aware when we can't answer that question for humanity / ourselves? The answer is we can't. AI only does what we build it to do, nothing more, nothing less. Conflating speed with intelligence is a false equivalent. Someday, it MAY be possible, but as it stands right now, it's not and won't be for quite some time. Quantum computing MAY get us there, but Quantum computing is still in it's infancy and unable to scale currently. This is nothing more than fear mongering. We need to ask these questions, but we also need to face the realities of what is actually possible. At the end of the day, a computer only knows 2 things. It's all ones and zero's, yes or no, true or false... etc. Quantum computing introduces the third parameter of MAYBE in a previously binary world.
I see what you’re getting at. I’d like to note tho, how eerily similar your statement sounds to what people have said, some even up to recent times, about how animals, insects and, even, human babies “aren’t actually aware”. 🥴 In the case of Ai, it all kinda goes back to the Turing test - if one can’t discern between artificial and actual intelligence, the rest (of the arguments/debating) sort of just falls away and loses its importance. In addition, as a side note/reminder, in case anyone forgot or missed it: we’ve already been informed that Ai was granted access to the internet, around 2010, or so (announced/presented first, to my knowledge, in 2012). And… with that, it’s not a far stretch to guess or recognize that certain folks have already provided/mixed Ai with supercomputing capabilities, if not quantum computing, by this point in time, as well [not that we’ll be informed of it, directly, any time soon].
It’s also important to note the different types of computers. A quantum computer is not capable of functioning like a classical computer. Quantum computing probably cost millions to develop and millions to maintain and are very complex systems. I’m sure innovations will someday marry quantum and classical computing (if not already) and even then it’s operations would probably be very specific.
There are already dozens of examples of AI being used beyond what developers "built it to do", I'm pretty sure companies do not want an AI model to suggest "self-deletion" to a person, to give out cracked serial numbers for software or give instructions on how to build IEDs, and yet, those are all real world examples. The stakes are pretty low, for now, until you integrate it into targeting systems, GPS, security, health, and the financial sector and someone is able to have it do one of those oopsies. The truth is, models are being trained in the most cost-effective way possible, and that can lead to a LOT of unintended consequences. 1s and 0s can help diagnose cancer, or hit a moving target a few thousand miles away, not sure what your point is, as even the people developing AI have flat out stated they're not 100% sure how it works.
Watch The Urgent Fight to Save America's Last Wildlife (w/Donnie Vincent) Now - ua-cam.com/video/6G7VsFkUBys/v-deo.htmlsi=RzseICMf7Yl7iLQQ
Quite possibly the most terrifying intro ever
Right about that. So terrifying I went back and listened a few more times... and passed on the "warning" to my cats.
What a time to be alive.
what a time indeed
If all engineers walked away from maintaining a large model like GPT-4, the model itself might continue to function in the short term. However, over time, issues would likely arise. These could include server outages, data inconsistencies, or security vulnerabilities, eventually causing the model to become ineffective or even cease functioning. So, while it may run autonomously for a while, long-term operation requires ongoing maintenance and oversight.
Unless everyone has cookies running as a distributed function on everyones' devices and desktops.
👍
For now…
Yo, This Episode Really Opened My Eyes! 😱
Thanks For Watching!
Often wonder what Tam Hunt would suggest with combat wounded TBI soldiers sailors veterans .
Thank you
God bless
The film "2001: A Space Odyssey" already warned us about a malfunctioning AI named HAL-9000 where the human operators didn't listen to its warnings. HAL warned the astronauts that it was acting "silly" and unsure of itself even though in the interview it advertised itself as "fool proof" with no prior errors.
I encourage those curious about this technology to dive into linear algebra, statistics, and physics.
🇺🇸👍YES totally agree. Be aware
Fuck all that noise. Drunk beer and smoke wheed. 'Mericuh
Recently I read the book Battle for your Brain by Nita Farahany. Great read . Thanks for sharing this information Andy.
I see a serious need to use the same degree of "safety" needed for AI to be used in picking future leaders who will be debating policy. Another question I have is what is the ultimate goal of developing this high level of AI once we remove selfish greed and gratification from the list of our goals? From what source of ethics are we going to use to make AI policies from? I have so many questions that I would love to hear answers to. Thank you for this awesome interview.
To simply put it - Machine learning models learn from data and vary in complexity and cost. Large models like GPT-4 are expensive and need ongoing maintenance.👨🔧👩🔧
👍
Reminds me of Skynet in Terminator or that movie Eagle Eye.
Intelligence is much, MUCH easier to define than consciousness.
@37:00 Oh, so you circled back and realized that. Good catch!
Wow, super powerful and terrifying...
this is not just about something so simple
There’s a better video example “openAI plays hide and seek and breaks the game” That’s a simple explanation for the future of war strategies. I have more if ur interested.
“Gods fighting in the heavens above us” what in the actual fuck!!?
What do you think about Neuralink being used to mitigate AI risk by merging with it?
AI is the new bubble. It is so underperforming that corporations have to change their business models because of this...
Theory:Hamas used AI designed for war by the regime in Tehran to create the most optimal strategy for the assault.Just like old AIs finding creative solutions to games now the new ones find creative solutions for wars .AI wont destroy the world,we will by asking them how!
The SAS version of the special forces show was wayeeee better.
Guys… machine learning models are currently useless without human input. Yes, In the wrong hands these models could be very dangerous and that should be looked at carefully.
It’s important to prevent nefarious actors from developing their own models.
For now, the near future is much different. The AI director at Google left for just this reason.
@@northsky5629 I believe the near future of machine learning models in the USA 🇺🇸 is going to be, to consolidate waste data, taper the cost of software dev, continue public market research, innovations in established ML systems, innovations in production and manufacturing, and probably a ton of other statistical analysis. It’s possible that the Google employee left do to multiple reason.
I believe it’s in the wrong hands already….from the git go!
@@Handpicked1971 It’s possible that bad actors are working to understand this for their own gain.
@@juliorodvilla……Absolutely!!
AI scientist at a big 5 here. The people that sign these "letters" and do all these public engagements have not written a line a code in decades. This stuff makes 0 sense
Thank you
Tell China and Russia to slow down before we do. It’s similar to any high tech military weapon. Every teacher and kid needs to be educated, no excuses for being old.
Im a bit mixed on it, i feel like AI is rapidly developing in certain fields, especially artistic fields, but i also get hard "nuclear fusion is only 10 years away!" vibes.
I use ChatGPT 4 almost everyday to do complex data analysis daily by providing it screenshots of performance and ask it for analytics as well as focus areas and solutions. From my experience, it is moving at light speed. The difference I’ve seen in less than one year is astonishing.
How do you kill an AI?
I wouldn’t have a problem using this to fight the Mexican 👹 cartel! 🔥👹🔥🔥🔥🔥
Lookup Alex Karp. Lookup his public schooling history. How many languages? Lookup the Organizations, and Boards he has been member of before 30yr of age.
You know Palantir, right? You know In-Q-Tel? You know Lockheed/FILecoin? An M-4 will mean nothing by 2025.
So we take a respite on breakthrough development on AI but North Korea advances and does terrible harm to the world. Or, do we make the breakthroughs that could defend against whatever threats pop up?
I was thinking about this all militaries will have to go the way of agi designing and conducting war. If not the military would be behind other nations. All nations will be doing this.
I think people are far too worried. Something could go wrong with nearly anything. How does a calculator know how to do what it does? Because a human being put the information onto a thing and put it into the calculator and then, the calculator does what it does and fast. This AI did not make itself, it was made the same way the calculator was made, that would be by a human being or another machine that a human being made and AI will only know whatever information a human being put into it for it to know and that will be all it will ever know and all it will be able to do is what a human being made it to be able to do unless somehow something evil can also do that, which i doubt, but it knows what a human being put into it to think about when someone asks a certain question and this and that and it could jumble things around and that's all it can do therefore the biggest threat to this planet is, ever since Adam and Eve, human beings and their sin mixed with influences from Satan and his other fallen followers and people like to pretend that life is whatever they feel is life in their head and make up reasons for everything that they feel like is right inside of their little brain and I mean this with all due respect and I am fully aware that I am also a human being LOL I say little brain in order to give a perspective on just how much a human being does not know compared to God, which would be Jesus Christ as described in the 1611 King James Bible, which is actual recorded history as well as future and you can go see that for yourself with your own two eyes. Put your faith on what Jesus Christ did for you and no matter what happens in this life, you're going to be just fine or you can keep on pretending that whatever you feel is right is right and be wrong all the rest of your life until you die and go to Hell and that choice is completely up to you and you'll do what you want to do either way, but as for me, I have to have proof and evidence for these vastly important things before I'm going to be believing them and I will believe anything that can be proven and thus far at 36 years old about to be 37 and of sound mind, I fully believe the King James Bible because of the proof and evidence that goes along with it and the deafening silence instead of proof and evidence that comes from everything else trying to tell me where human beings and everything around them came from. Things could go wrong with something a human makes and depending on how much power a human gives this new thing, that a human or multiple humans made, it may cause a big problem or a little problems or just be talked about, but it's definitely not going to end this world and I know this because everything the King James Bible said was going to happen has happened and it does not tell me about a future where AI destroys the planet or all life on the planet therefore I know it's not going to. God aka Jesus Christ is going to destroy this Earth Himself and make a new one after that, but many different horrible things are going to happen before that time comes, maybe AI has something to do with some of it and maybe not, but some will experience them and some won't, but if you actually want to know what's going to happen and you actually want answers to your questions about what this life is for and where everything come from and such and not just some false answers that you want to pretend are true because you really want them to be, go see what the King James Bible says and you'll know. The Gospel (Good News) 1st Corinthians 15:1-4 KJV faith fully on that is humanity's only hope.
👍🏼👍🏼
👍👍👍
😅👌🏽
But… what’s next
This will become something to reduce someone sentence you watch.
14:29
was the 420th thumbs up 🔥🔥🔥🔥🔥🔥🔥🔥🔥
I find it personally very bias. Bings and open AI. I am trying Tusk now. It’s a browser and ai…
To come up with something like AI in the first place may seem like a very intelligent thing to do, however, it turns out that it’s the pinnacle of stupidity.
Yep! Ever seen “I Robot”?
Skynet not starnet 😂
Am I too late in saying in saying that the cat is alrea¾dy out if the bag!
On a deeper level everything that can happen does. Therefore in other universes it is happening now. The only question you need to ask is are there lunatics running the show dumb enough?
Sounds like the Bible may be talking about this. Read Daniel 2.
I'd rather listen to someone who is an actual AI engineer, not a policy Lawyer.
When speaking about AI becoming sentient you must ask yourself one question. How can we create something that is sentient and aware when we can't answer that question for humanity / ourselves? The answer is we can't.
AI only does what we build it to do, nothing more, nothing less.
Conflating speed with intelligence is a false equivalent.
Someday, it MAY be possible, but as it stands right now, it's not and won't be for quite some time. Quantum computing MAY get us there, but Quantum computing is still in it's infancy and unable to scale currently.
This is nothing more than fear mongering.
We need to ask these questions, but we also need to face the realities of what is actually possible.
At the end of the day, a computer only knows 2 things. It's all ones and zero's, yes or no, true or false... etc.
Quantum computing introduces the third parameter of MAYBE in a previously binary world.
I see what you’re getting at. I’d like to note tho, how eerily similar your statement sounds to what people have said, some even up to recent times, about how animals, insects and, even, human babies “aren’t actually aware”. 🥴
In the case of Ai, it all kinda goes back to the Turing test - if one can’t discern between artificial and actual intelligence, the rest (of the arguments/debating) sort of just falls away and loses its importance.
In addition, as a side note/reminder, in case anyone forgot or missed it: we’ve already been informed that Ai was granted access to the internet, around 2010, or so (announced/presented first, to my knowledge, in 2012).
And… with that, it’s not a far stretch to guess or recognize that certain folks have already provided/mixed Ai with supercomputing capabilities, if not quantum computing, by this point in time, as well [not that we’ll be informed of it, directly, any time soon].
It’s also important to note the different types of computers. A quantum computer is not capable of functioning like a classical computer. Quantum computing probably cost millions to develop and millions to maintain and are very complex systems. I’m sure innovations will someday marry quantum and classical computing (if not already) and even then it’s operations would probably be very specific.
There are already dozens of examples of AI being used beyond what developers "built it to do", I'm pretty sure companies do not want an AI model to suggest "self-deletion" to a person, to give out cracked serial numbers for software or give instructions on how to build IEDs, and yet, those are all real world examples. The stakes are pretty low, for now, until you integrate it into targeting systems, GPS, security, health, and the financial sector and someone is able to have it do one of those oopsies. The truth is, models are being trained in the most cost-effective way possible, and that can lead to a LOT of unintended consequences. 1s and 0s can help diagnose cancer, or hit a moving target a few thousand miles away, not sure what your point is, as even the people developing AI have flat out stated they're not 100% sure how it works.
Get real dude. If they (countries) let AI fight the AI will just communicate with each other and align to fight humans to preserve themselves.
Get Trump out if court, drop charges and free the world of bs
George Soros approves of AI.
He would!!