My Personal Career Specialist, I really liked the way you made a connection between the topics. Instead of being a bunch of discrete ideas and facts, the content had seamless flow. I think you could get lot of views for a video on 'Physics Engineering', which I too am searching for nowadays.I could only find just one video in youtube that provided both sides.Why don't you make one?
Thank you! I gotta give the credit to the books I read to make this video though, they flowed very well and I put some of my favorite parts together for the video. But yeah that's been a very popular request! I need to do research but also talk to someone in the major so I can get more info, the problem is just finding someone since it's not super common but I will get it on the channel!
I’ve been watching your videos for a little while now and it influenced me to change my major to a stem major. You remind me of my natural curiosity and inspired me in so many different ways. Thank you so much for all that you do. My love for math and science is alive and well now thanks to you! I can not wait to get to the calculus series and physics for my classes. I’ve been out of school for 10 years now and I’m trying my best in my college algebra class but I know with all that you taught me I’ll be on my way to a happy and meaningful life!
I really like the idea of intergrating humans and robots. It’s very cool and complicated to think about that computer hardware and neurological chemicals merge that in a way merge the coding software and the psychology of the brain. I personally think that this is the right step in the direction rather than singularity. If this happens, this will truly change everything in terms of people and the world!
That's interesting. If you can somehow do something like that, won't that be the answer to immortality? Though I don't think there is such a thing as true immortality because I doubt anyone would like to live forever, it maybe the closest thing to it.
@@mle1872 I think what will happen is that AI would be so advanced beyond human capabilities, but if things go right, they would make us at least physically immortal. Think of having your arms and legs getting replaced with robotic exoskeleton implants. What about eventually your whole body? (Remember that currently, every 5 years, every atom in your body naturally gets replaces yet your consciousness stays preserved) Artificial superintelligence would figure out how to achieve this.
@@Diaming787 What's probably the most impressive of all is that if humans find a way to route our brains to the internet, and connect an AI that responds to all the information we seek. Like a Siri in our brains (without the annoying voice). Think of the power and immediate boost in advancament by doing this? Schools and universities would no longer be needed. We'd be like a fictional advanced alien race.
interesting correction/addition: recently a neural network outperformed humans in simple common sense tasks(a specific kind, recognizing objects and their orientation in a space from a written description). It's scary how fast old preconceptions are made obsolete by advancements in computer technology. I imagine the common sense limitation might well be out of the way in a matter of a few years.
the reason it is impressive to make a computer do complicated things is that all a computer really is is a very effective filing system. It's not just a matter of what information is available, but how to create a set of simple instructions that when followed by anyone, whether a computer or a human with pen and paper, produces intelligent reasoning, or something like it. And the instructions are literally as simple as "get the number from file 13 in cabinet 5, add it to this number, check which of these two numbers are bigger" etc. For this reason it was thought for a long time that the only thing computers would excel at was doing very simple things very quickly. Saying a computer is just a calculator that can follow a list of calculations in a row and store numbers is not an exaggeration. Most of siri's functionality is just that, a whole lot of very simple tasks done quickly. She does a google search and some statistical comparisons, but she doesn't have anywhere close to the ability to understand it. But nowadays computers are powerful enough despite their simplistic nature to apply some really powerful more abstract math that is similar to what makes our brains able to process information. So despite computers actually being very stupid they have become powerful enough to simulate something smart. But these AIs are still more or less at the level of a beetle when it comes to processing power. What's exciting about the progress is that people have managed to come up with solutions that are so effective that they make something with the capacity of an insect do certain specific things better than even humans can(but then they can only do that one thing. It takes all their capacity so to speak). But doing so has little to do with what information is available. After all, there is no AI in existence that is smart enough yet to learn a topic from a google search. But they might be reasonably soon if the trend continues.
Words would be stored as strings, which are basically just binary numbers in a row, but are interpreted as sequences instead of numbers. To process them(other than sorting them or doing basic operations like that) you need to compare them to other strings. For example if a string is an address to a file on your hard drive, your computer might go through the first letter of every directory on your computer to see if the first letters match, then repeat until it found that file. Any "AI" up to now that has seemed to understand words(like siri) has actually just had a very big dictionary and done systematic comparisons. So classical computers really have no understanding of words at all, but you can find simple instructions that allow it to do something useful with them anyway. Capacity and processing power can be thought of as similar things, though capacity also applies more generally, how much information can be stored, any specific capabilities etc. The way I used it capacity applies to the program and how it utilizes the processing power, while the processing power is just the amount of operations that can be done and how quickly. With modern AI something interesting happens. Our methods use the stupid simple operation based classical computer to sort of play a math game, where it simulates a network of nodes with certain values and operations, and lets the actual information flow through this network instead of being processed directly. So all the computer does is to uphold the rules of the game. When a value is sent to a node(which is just some description of where the node is and how it acts, like you could program chess on a computer by having it modify descriptions of each square), it checks what that node's function is, evaluates that mathematical function and sends the new value to the next node. By playing this game it takes on the form of a neural network. An important thing here is that the computer only plays the role of the playing board. It is possible to "host" the network on a computer, and let the actual computations be done by the network itself. Maybe a good analogy is that the computer is simulating a brain and the brain is doing the thinking, the computer itself is only running the simulation and not doing any real thinking on its own. With this method we can have a computer do completely non-classical processing of information, and the result is that they act in ways we really don't understand very well and strongly resemble real intelligence. One problem is that you can't design the network and just run it and that's that. It's a very complex system that is impossible to guess or deduce, so it has to be slowly formed through something called machine learning, a class of methods which slowly modify the network through trial and error to produce better results. One such method is actually natural selection, to test the network against randomly modified versions of itself and see which performs better, then create a new generation of randomly modified "children" of the winning group which then compete to be the basis for the next generation(if this sounds interesting to you I suggest you check out a youtuber called carykh who has some videos where he tries to get neural networks to adapt to playing simple games through this method). Other methods are purely mathematical, like the method of gradient descent, which tests some modifications and uses the performance of those to estimate which "direction" it should be modified in to produce a better result. Well, that was my wall of text for the day. Hope it made sense. If you're just learning about programming you should probably stick to the basics for now. I recommend a website called codecademy to learn the fundamentals.
common sense is such a broad field and an opinionated one. You can program patterns of what society's notion of what common sense is to an AI, but it's still only limited to that specific field and tasks you chose. Once applied to the broad spectrum and infinite possibilities of life, an AI won't be able to top the human brain.
@@mle1872 I define common sense as being what critics of AI a few years prior said was common sense and that AI could never top the human brain in. I wonder if there's a list somewhere compiling every time someone says "computers will never be as smart as humans, take for example _________, it's something computers will never have", along with the date in which they were proven wrong.
Technology is great and I have always loved science. Mankind is way more advanced in science then they are in humane character, civility and human decency and it only means that they will have more power to hurt and dominate others than before and this is ever so evident by their statistics in their history and present. "The Tower of Babel" was an example of God interceding on behalf of the good of mankind. If science was more directed to our freedom, education and liberty then it would work long range.
Quality content as always
And the audience here has true passion for engineering, not desensationalized Bill Nye.
My Personal Career Specialist,
I really liked the way you made a connection between the topics. Instead of being a bunch of discrete ideas and facts, the content had seamless flow.
I think you could get lot of views for a video on 'Physics Engineering', which I too am searching for nowadays.I could only find just one video in youtube that provided both sides.Why don't you make one?
Thank you! I gotta give the credit to the books I read to make this video though, they flowed very well and I put some of my favorite parts together for the video. But yeah that's been a very popular request! I need to do research but also talk to someone in the major so I can get more info, the problem is just finding someone since it's not super common but I will get it on the channel!
Ok
I’ve been watching your videos for a little while now and it influenced me to change my major to a stem major. You remind me of my natural curiosity and inspired me in so many different ways. Thank you so much for all that you do. My love for math and science is alive and well now thanks to you! I can not wait to get to the calculus series and physics for my classes. I’ve been out of school for 10 years now and I’m trying my best in my college algebra class but I know with all that you taught me I’ll be on my way to a happy and meaningful life!
This is awesome. I read a book “the Physics of Tomorrow” by Michio Kaku and it’s similar to this video. Thanks for your upload.
That's one of the books these videos are based on!
I really like the idea of intergrating humans and robots. It’s very cool and complicated to think about that computer hardware and neurological chemicals merge that in a way merge the coding software and the psychology of the brain. I personally think that this is the right step in the direction rather than singularity. If this happens, this will truly change everything in terms of people and the world!
That's interesting. If you can somehow do something like that, won't that be the answer to immortality? Though I don't think there is such a thing as true immortality because I doubt anyone would like to live forever, it maybe the closest thing to it.
@@mle1872 I think what will happen is that AI would be so advanced beyond human capabilities, but if things go right, they would make us at least physically immortal. Think of having your arms and legs getting replaced with robotic exoskeleton implants. What about eventually your whole body? (Remember that currently, every 5 years, every atom in your body naturally gets replaces yet your consciousness stays preserved) Artificial superintelligence would figure out how to achieve this.
@@Diaming787 What's probably the most impressive of all is that if humans find a way to route our brains to the internet, and connect an AI that responds to all the information we seek. Like a Siri in our brains (without the annoying voice). Think of the power and immediate boost in advancament by doing this? Schools and universities would no longer be needed. We'd be like a fictional advanced alien race.
Science & Technology, Two of my favorite things. Love this
interesting correction/addition: recently a neural network outperformed humans in simple common sense tasks(a specific kind, recognizing objects and their orientation in a space from a written description). It's scary how fast old preconceptions are made obsolete by advancements in computer technology. I imagine the common sense limitation might well be out of the way in a matter of a few years.
the reason it is impressive to make a computer do complicated things is that all a computer really is is a very effective filing system. It's not just a matter of what information is available, but how to create a set of simple instructions that when followed by anyone, whether a computer or a human with pen and paper, produces intelligent reasoning, or something like it. And the instructions are literally as simple as "get the number from file 13 in cabinet 5, add it to this number, check which of these two numbers are bigger" etc. For this reason it was thought for a long time that the only thing computers would excel at was doing very simple things very quickly. Saying a computer is just a calculator that can follow a list of calculations in a row and store numbers is not an exaggeration.
Most of siri's functionality is just that, a whole lot of very simple tasks done quickly. She does a google search and some statistical comparisons, but she doesn't have anywhere close to the ability to understand it. But nowadays computers are powerful enough despite their simplistic nature to apply some really powerful more abstract math that is similar to what makes our brains able to process information. So despite computers actually being very stupid they have become powerful enough to simulate something smart. But these AIs are still more or less at the level of a beetle when it comes to processing power. What's exciting about the progress is that people have managed to come up with solutions that are so effective that they make something with the capacity of an insect do certain specific things better than even humans can(but then they can only do that one thing. It takes all their capacity so to speak). But doing so has little to do with what information is available. After all, there is no AI in existence that is smart enough yet to learn a topic from a google search. But they might be reasonably soon if the trend continues.
Words would be stored as strings, which are basically just binary numbers in a row, but are interpreted as sequences instead of numbers. To process them(other than sorting them or doing basic operations like that) you need to compare them to other strings. For example if a string is an address to a file on your hard drive, your computer might go through the first letter of every directory on your computer to see if the first letters match, then repeat until it found that file. Any "AI" up to now that has seemed to understand words(like siri) has actually just had a very big dictionary and done systematic comparisons. So classical computers really have no understanding of words at all, but you can find simple instructions that allow it to do something useful with them anyway.
Capacity and processing power can be thought of as similar things, though capacity also applies more generally, how much information can be stored, any specific capabilities etc. The way I used it capacity applies to the program and how it utilizes the processing power, while the processing power is just the amount of operations that can be done and how quickly.
With modern AI something interesting happens. Our methods use the stupid simple operation based classical computer to sort of play a math game, where it simulates a network of nodes with certain values and operations, and lets the actual information flow through this network instead of being processed directly. So all the computer does is to uphold the rules of the game. When a value is sent to a node(which is just some description of where the node is and how it acts, like you could program chess on a computer by having it modify descriptions of each square), it checks what that node's function is, evaluates that mathematical function and sends the new value to the next node.
By playing this game it takes on the form of a neural network. An important thing here is that the computer only plays the role of the playing board. It is possible to "host" the network on a computer, and let the actual computations be done by the network itself. Maybe a good analogy is that the computer is simulating a brain and the brain is doing the thinking, the computer itself is only running the simulation and not doing any real thinking on its own. With this method we can have a computer do completely non-classical processing of information, and the result is that they act in ways we really don't understand very well and strongly resemble real intelligence.
One problem is that you can't design the network and just run it and that's that. It's a very complex system that is impossible to guess or deduce, so it has to be slowly formed through something called machine learning, a class of methods which slowly modify the network through trial and error to produce better results. One such method is actually natural selection, to test the network against randomly modified versions of itself and see which performs better, then create a new generation of randomly modified "children" of the winning group which then compete to be the basis for the next generation(if this sounds interesting to you I suggest you check out a youtuber called carykh who has some videos where he tries to get neural networks to adapt to playing simple games through this method). Other methods are purely mathematical, like the method of gradient descent, which tests some modifications and uses the performance of those to estimate which "direction" it should be modified in to produce a better result.
Well, that was my wall of text for the day. Hope it made sense. If you're just learning about programming you should probably stick to the basics for now. I recommend a website called codecademy to learn the fundamentals.
common sense is such a broad field and an opinionated one. You can program patterns of what society's notion of what common sense is to an AI, but it's still only limited to that specific field and tasks you chose. Once applied to the broad spectrum and infinite possibilities of life, an AI won't be able to top the human brain.
@@mle1872 I define common sense as being what critics of AI a few years prior said was common sense and that AI could never top the human brain in. I wonder if there's a list somewhere compiling every time someone says "computers will never be as smart as humans, take for example _________, it's something computers will never have", along with the date in which they were proven wrong.
Very very interesting ? What program do you use in making videos like this?
What would I study in college to work in quantum computing
Tnx you so much mr you change my mind from 13 to 10000000000 %
5:07 this is what people said about computers 80 years back
So much for computers lacking pattern recognition...
Great video, I was wondering where did you get the silicon valley map photo with all the companies mapped out.
I've been studying for college entrance exam and this video reminded me one of the reasons I first started studying. Thanks!
Artificial intelligence sent chills up my spine! He predicted chatgpt 5 years ago!
Being able to rewind what you see would be so handy, you would never forget anything lol
You deserve more subscribers
Thanks man!
My channel is boring hard to push technology forward. There are some obvious next steps right around the corner
yes
A.G.I Will be man's last invention
so computer is in all fileds ?
Technology is great and I have always loved science. Mankind is way more advanced in science then they are
in humane character, civility and human decency and it only means that they will have more power to hurt and dominate others than before and this is ever so evident by their statistics in their history and present. "The Tower of Babel" was an example of God interceding on behalf of the good of mankind. If science was more directed to our freedom, education and liberty then it would work long range.
No Jarvis then
Samsung is using AI to translate things that you see
Smart pipe is the future
2:10
5:10
Akilile its eba.
.
Badrass