Next Step in Evolution - "AI Domination" | Singularity | AGI
Imagine a time where machines are working millions of times faster than us A time where AI is so strong that it can fulfill all your dreams But it's not necessary that all those dreams are good, they can be nightmares too So today we are going to talk about Singularity What is Singularity and will it be achievable or not? So first of all, what is Singularity? Singularity is a state where AI becomes so fast that it starts developing itself And the technological advancement increases infinitely Today we see that a phone launches an iPhone in a year Next year it becomes 20% faster Imagine if the technology becomes 1000 times faster in the next day, then this is Singularity Singularity is a hypothetical time where machines are able to upgrade themselves To upgrade themselves, they are inventing new technologies Many people say that this is a utopia movement where we will get whatever we want We will be lying at home and ask AI to bring a burger for us We will get whatever we want Many people say that AI will go out of our control and will kill humans And many people say that this is not possible So now a days, there is a research on Singularity Actually, no one can predict this If it will happen, then in which direction it will go, only time will tell Now let's understand the evolution of AI In 1943, the first paper was published where the neural engine was discussed with logical gates Basically, how can we connect our brain with gates If you have studied computer science, then you must have studied gates and OR Which had an output of 0 or 1 Basically, our brain is working in the same way So this concept was discussed in that paper In 1950, a machine was tested with a human And there was a human evaluator who was checking the machine and human In 1956, the first AI conference was held And in 1966, the first computer program was made Which worked on neural language programming Which was a command and it used to reply. It was named ELIZA. Maybe you have used it if you are my generation I am not that old but I remember that there was a DOS file named ELIZA We used to double click it and ask questions. It used to give us one level of answers In 1974, AI was used in a good way Here, it was worked on uncertainty Till now, it was like this is the answer and this is the reply But if a new condition is created, then you have to find the answer It was worked on that So basically, this technology was used for the research of bacterial infection In 1985, a big thing was done for machine learning An algorithm was made which can train itself Basically, if you talk about AI, then it has to be trained The training mechanism was started here In 1977, a big thing was done An AI computer, IBM computer defeated a chess champion Here, it felt that chess will end and AI will be used and people's visibility will be reduced In 2011, IBM's Watson system participated in a television game show and defeated the old champion In 2012, a big thing was done in image processing To win the image net challenge, Alex net made a model of image classification Where AI can organize the image In 2015, there was a product of Google named AlphaGo It defeated a Go champion Go is a very complex game It was believed that it is very difficult to train AI In 2018, a big thing was done It was GPT. GPT-2 and it changed the entire system of neural language processing People never thought that it will go in this direction And then chat GPT-3 came Now 4 came This is how AI is evolving Till now, you have heard about chat GPT An image model is also running on which many people are working Chat GPT's parent company OpenAI is working on stable diffusion Stable diffusion is a different company and using its technology, it is making its product Many other players are making their own It makes an image according to the text you write At present, it is not sophisticated enough to reach the level of chat GPT But according to me, you can do anything on chat GPT But the story is that everyone is afraid of chat GPT-5 That it will not be made Now they have to stop it Because if it is made, it will take AI to a singularity Basically, everyone is afraid because if AI is given so much power Then AI will become so strong that machines can talk to each other This is not a new thing This has happened before also Machines have developed their language and have been caught while talking to each other So if this happens, that machines are talking to each other And that next level is evolving Then we will see singularity very soon AI will become so strong that it can do anything Now you tell me, do you understand the concept of how AI has been developed? You saw that big game players were defeated And this is a very old thing Don't you see chess today? Do you see AI chess today? No, right? So many people are afraid that AI will come and do this and that See, understand AI is a very old thing and not a new thing If I talk about calculator, then what is calculator? To make my calculation easy 2 plus 2 into 5 into 7 into this into that It is calculating very fast Machine is faster than me We never said that calculator will replace us Because we have adopted that thing in our life So all the new technologies that are coming Because the problem is that the first AI that was running was suppressed People used to use it but didn't know Today if we use Google assistant or Siri or any voice assistant That is also AI If we talk about cars, self-driving cars That is also AI Self-driving is very advanced Nowadays, sensors are also coming in normal cars It tells you that this is coming on this side and that is coming on that side Take care of yourself I am talking about my engineering time We made a line follower It used to follow a line That was also a type of AI We trained it that you have to walk like this and like this So why are people scared now? Because chat GPT is very popular now And it is going to hit a very big industry And that is according to me The whole knowledge industry I have been against this since childhood I don't understand this what is what is what is problem What happened in this year, what happened in this year, who did what, why did what If it is not relevant for me, then we should learn history But if it is not relevant for me, then I don't need to know that Nowadays, if a calculator is calculating for me, then I don't need to go into that level of calculation We are being taught trigonometry, if we don't want to go in science field, army, air force Where trigonometry is not required, then why should I study? So if I talk about chat GPT type language processing models Then it is going to hit that industry Not now, but in future And on the other side, image processing can ruin the whole graphing industry But now I think it is not evolving to that level Where you can generate text to image of that level So if in future AI evolves to that level and makes my work easy for me Then I don't understand what is the problem in that So you got to know that there are two three types of vision One is optimistic and one is pessimistic Optimistic is saying that if AI becomes strong, then utopia will be created Some people say that if AI is created, then it will rule us It will make us a servant, it will do this and that I don't think so Rest, I am just guessing And everyone is guessing If you say in the comment section that it is like this, it is like that It is your guess No one knows in which direction, if singularity is achieved, then where it will go Now let's understand what are the drawbacks of AI We have heard many pros that our work will be easy Technological advancement will happen If we have to make any technology, then it will take 2 years to make it Maybe we can make it in few hours In today's time, when we made vaccine during covid, AI was involved in that too In future, there will be many medicines that AI can make and give us Simple thing, we have limited knowledge in our past A doctor has studied so much knowledge AI knows all the knowledge of history If we train it on that level Then it can use all that knowledge to make new conclusions to think So by doing calculation very fast, we can get all the answers But there are drawbacks also If I use calculator a lot, then my calculation gets spoiled A person takes time to think but a computer will not take time If we talk about AI, AI comes at the next level So first thing is that the features of AI will be known to limited people Mask will not be known If I talk about chat GPT, then most of the people don't know about it And those who know about it, they ask me how to do it And those who are bringing it in practical business use cases, think how good they are doing their work So here a big inequality barrier will be created Those who have AI knowledge and can use it well, they will be class apart Prompting is a very important skill I think in future, it will not be relevant to our studies Prompt engineering will be there How to search and tell AI that this thing should be done Now there is a negative side to it, AI is strong and powerful Now we are talking about it in positive direction If someone starts training it in negative direction, then it can be misused To hack people, all our devices are technology linked If AI wants to hack something, it can do next level hacking It can provide security but can also do hacking One thing that many people are afraid of, I make business news and tell people that there is no need to be afraid of AI People say that job will be lost, yes it will be lost Many jobs will be lost It is possible that in the next 10 years, many jobs will be irrelevant I am not talking about India, I am talking about global level I think there will be many places in India where internet will not be of the level we are thinking But again, think about it, there will be more inequality AI will be a different person for them AI will be there in their phone They are asking them questions and they are answering them One big problem is that if we are dependent on AI, then it will also have bias Some people say that AI can't be biased, it is getting core knowledge We have talked about it, AI has to be trained So you have to give something to train AI If data is biased, then it will have its own opinions If opinions are made, then there will be some bias One more problem is that AI is seen as a hallucinator What does hallucination mean? You ask him some questions and he is answering them with conviction So this is also a drawback, if we are dependent on AI I asked who is the father of nation, he said Bhagat Singh I said if AI said so, it is right Because we don't know what is happening in the future I am taking a vague example, it is possible that this will be taught in the textbook So let's talk about the future of AI I am talking about the immediate future, not the singularity future I am talking about what is currently possible First of all, we are seeing GPT growing GPT is a model where we ask questions and it answers them I think prompt engineering will increase And people will do a lot of work with GPT I am using it for my work since last 15-20 days And I am planning to make a micro course on it A cheap course and a quick course And you will learn prompting and how to do it We were doing a test and I asked AI I have a daughter, so what should I name her? Basically, the name is taken from the pandit It told me that So basically, we name the child I am talking about my culture After 40 days, the name is given Now when the child is in the hospital, the name is written and the name is given in the birth certificate And if you don't write it, then you have to update it later If I know the name of my child on the first day itself Without going to a pandit I will write it there After 40 days, I will do what I want to do And then I will use the official name It is so easy to see all these things And I asked a next level question Where we are talking about Hindu rituals If we talk about GPT, it should have been internationally trained And it is according to the India model I can write content in English But the starting of this video was made in Hinglish And chat GPT helped me a lot in researching this video Second, we can see the home system is improving The lights in your house are going towards AI Now automation is not there yet You have to tell the voice assistant I think voice assistant will be so powerful Basically, you are seeing the level of chat GPT All these things will be added in the assistant And you can make it do next level things Now a days, there is a small drawback of chat GPT You have to do a lot of things Do this, do that If a system comes which will do everything by itself And it will search everything by itself And will bring the answer to you You will think this is not possible But this is possible now Recently I saw a product called baby AGI It is open source And you can ask questions to it It will make you a task And will bring you the result I am asking it to tell me what is this It is searching or I am asking it to make a script for this video To make a script, it is searching the topic It is searching the outline It is researching about it If we talk about chat GPT, there is a small limitation Which chat GPT has recently removed They have removed the plugins We will see one thing immediately A lot of companies will link with chat GPT And they will connect with the plugins I will ask chat GPT to order tomatoes from here And it will be done very soon I was reading about Shopify, they have developed this plugin Next is image processing Because the game of image processing is limited Everything is getting developed So I think image processing will go to the next level I am saying that you will give command to AI And it will give you an image which will look real In AI, sometimes you can see something in the finger And there are some errors It is not perfect You can't say it is 100% perfect Sometimes it is not perfect Or you have to work hard Again this is a science How to use image processing model And bring the result If we talk about chat GPT We write the command and it gives the result We don't have to work hard on the prompt But if we talk about image processing We write the prompt and it shows the image Then we modify it Then we write something else regarding it Then we write something else And we build the environment Next is coding And I think coding will give a big shock to AI system coders Because even now if we talk about chat GPT It is coding itself And this is just the beginning Microsoft has announced another product It is out Which will be able to do next level coding Let's see how it will evolve over the period of time If we talk about jobs, then jobs can go to the image And it can go to the level of monotonous work 1 2 3 4 1 2 3 4 This work has no sense If you do this work, then please change it If you think AI can replace you Where your brain is not there AI doesn't have brain We are talking about singularity AI has the ability to think But now we are talking about It tells us when we tell it It is not doing everything by itself If we talk about music generation I think it can happen immediately In 1 to 1.5 years, there are chances
That we will hear AI related songs Singers are singing for AI They are not singing for themselves AI is singing for them AI is very useful in research and technology To find medicines or to advance technology AI is helping a lot It is helping now and will help more Content creators are also in danger If I talk about it If you are prompted, then you don't need to watch videos What is singularity If you want to know about it's impact What is the latest There are no chances that you will get it But if you want to know about case studies There are many cases on intellectual induce AI has started developing I thought it will be replaced But after sometime I realized that Because I believe that content should be made Which is irreplaceable in future If someone will replace me in future Then it doesn't make sense for me If you want to see case studies You can directly search from chat gpt What is the story of this company And what can I learn from this You don't need to watch 15 minutes video It gives you straight answers The way chess players Are not replaced by AI Content creators are not replaced Because people like people more I am not talking about AI I am telling you this because I think This can create a revolution Brain chips And it will link you to the world of computer What is the difference between computer and us? Our capabilities are limited You can make computer unlimited You can use any number of ram, power, cpus Any number of cpus You can make a station where only computers are there And it is giving you power So brain chips Will play a big role in future It will create imbalance But I think it will be interesting You have connected to the computer We are talking about prompting So there is no need of prompting I thought it should do I never thought it should be hungry It knows that i am hungry I think it can be next level But again if you use brain chips, it can be a hack It can hack your brain Yes, this world is far away There are no human trials One more thing If we want to go to next level, we need to make more super computers And super computers are of cheaper grade Because they are cheap and super computers are powerful We need to go to that grade Because AI needs more computing power AI is working at this level If you use chat gpt You will give a big command to it But now there are very small capabilities If you want to take it to next level You need more super computers So we need a big switch in computer technology I think It should happen soon And i think soon AI engines will be made AI cores will be different Right now, your computer is good for basic processing But not trained for AI We need privacy and AI So we can train AI in our local level But right now We don't have hardware for that So i think there is a gap And opportunity So i think singularity Will not be achieved soon And if it will be achieved, it will be a blessing for humans It will be a good thing This is my point of view, you can share your point of view in the comment section And one small appeal Please focus on AIU Because if you don't know AI, you will get obsolete very soon This is what i am saying There will be a big change in AI Don't be on the lower grade Learn as much as you can There is a lot available on the internet Or you can practice And with this, today's video ends here I hope you liked it Bye Good night Good morning whenever you are watching this video.
2023-04-15 13:08
