First off, thanks for this comment. Its the kind of discussion I come to this site for.
I wish the show host hadn't interrupted so much with baseless "what ifs."
Its typical Nino strategy. He always plays devil's advocate, on behalf of this more black pilled listeners, and it always works. Right now, we have too many patriots who have let the pendulum swing too far to the right, so to speak, when it comes to anything "digital", esp after the Plandemic days. So, this kind of devil's advocacy makes them feel like they are being heard, and their big fears are being addressed.
He did the same thing during "46" years, always complaining to the point of sounding blackpilled himself, while letting himself be talked out of that despondency by the various people he interviewed.
There is also the aspect where he cuts of juicy bits so he can discuss them on his payed channel, often using youtube censorship as excuse. I think in this video there wasn't a whole lot of that just one or two instances.
I have studied AI, and I have found it seriously lacking at present. It will have to develop a lot in the next ten years to even be usable for simple things I would want to do with it. I tested ChatGPT a while back with simple tests that it failed completely.
I hate using the word AI as much as possible, so I will use LLM (large language models) because for firstly thats what they are - simple language models, and secondly, it reminds us that there is no "intelligence" and they are not "beings" but just tools.
So the utility of LLMs depends vastly on how you use them. They canbe extremely powerful if wielded correctly, but extremely disappointing if wielded wrongly.
Imo, wielding LLMs correctly requires us to decide which part of the hard work we want to unload on them, and which part of the work- esp thinking aspects - we want to do ourselves.
For instance, not withstanding the fact that it offered your Uncle's name, I would argue that LLMs can be very powerful in scraping genealogy data, compiling them, deducing connections through large number of clues that might be hard to manually analyse, and then build a system that would provide the answers, rather than using directly to find answers.
LLMs are also great at obtaining a list of very relevant source material for nay research, summarize them, prioritize them, while the human does the actual research.
I have been a coder for 35 years, and my entire perspective of software development has completely changed in the past 3 weeks after GPT-5 was released. I can now build complex software in minutes and I haven't written a single line of code in the past 3 weeks while i have build 20+ apps for myself, and even though I am a hardcore coder, I am enjoying this experience.
And the growth of LLMs is exponentially faster than any other technology. The day I first accessed the internet in early 90s, to the day it became a somewhat useful tool for me, was almost 4 years (even though I used to be engrossed in the early days of internet daily). And almost 10 years before it became a tool that non technical people can use.
Compare that to the day Meta released their first LLM - April of this year. And in 6 months, it has completely changed the way I have been doing a core part of my life for 35 years.
I want an AI that can search through the contents of over 100,000 books and answer any questions I have.
This is one thing I want as well. They have something called RAGs right now and I set one up, but they dont work as well right now. But the quality of LLMs, and esp the reasoning LLMs is growing so fast that I think in the next 2 months I will be able to build a system like this that would work better than a human. If I do it, I will definitely ping you.
I want an AI that operates solely in my office with no connection to the outside. Okay, I just rambled too much here, so I'll go on.
This is my personal passion as well. For the longest time (since April that is) I have been using local LLMs running on my own Nvidia to do my work. But unfortunately the computing power is too little and the quality of these LLMs is very limited, when you compare it to GPT-4 or GPT-5, for instance.
So I have shelved this passion for now, and embraced GPT-5. But I know that as the math behind LLMs become advanced, and the GPU hardware tech keeps advancing, pretty soon we will be able to run GPT-4/GPT-5 level LLMs locally.
I think that would be really necessary for companies working on sensitive information to be able to start using LLMs well.
Only gold and silver are real money.
Depending on how you define "real money" to be, if you use an objective definition that money is something that has inherent value, then only a quantum of energy or a quantum of computational power will qualify as "real money". These two have the power to materially transform things, and hence have inherent value.
But if we define money to be the something that people are willing to exchange for something of real value - still a subjective definition - gold and silver are by far the closest to real money
The problem is not even that. The real problem is that money has stopped being "thing of value" and has become "debt".
The "zero proof" stuff seems like a good idea. But there still has to be a "someone" somewhere who sets it up and controls it. It has to reside on computer servers somewhere.
This is the beauty of "Zero Knowldge Proof" actually. You dont need an external computer server to carry out the transaction. If I were to want to verify your age, firstly, you will only exchange data with me. Neither of us will need to exchange data with any other server.
You will have a encrypted certificate on your device given to you by the authority that issued your ID. I will have an encrypted key of the same authority on my device.
I will use my key to validate a piece of data generated by your device using your certificate and thats all. There is absolutely no necessity to talk to any external computer.
Of course, some people may choose to store their private certificates on a cloud server, but thats like people leaving their bitcoins on the exchanges. Its a stupid thing to do, but people are free to be stupid.
The very oldest people don't have the background knowledge to understand tech, and the very youngest are too stupid to care about tech,
This is allthe more reason why we need to put in a trustless decentralized system in place, so that future generation does not need to worry about it.
But making all this work ina. user friendly way while still keeping it secure is a challenge that comes in at implementation phase. What I am really concerned is for the digital warriors to always watch like hawks on any implementation of these digital ID systems, to ensure they are zero proof, not centralized and protects privacy at source - not sending any private data to any servers, period.
I refer to "real money" as what the US Constitution says. Congress was supposed to "coin money," but they gave that job to the Fed, and then started doing a bunch of other stuff that's unconstitutional. Also, gold has always been money for thousands of years. It's physical and fungible. Anyone can understand it. I'm still on the fence about blockchain money. That's another subject I need to research, if I just had more hours in the day.
I will be glad when I can have a computer system at home that has the horsepower, memory, speed, and storage to do what I want. I built my current computer 13 years ago. It was state of the art then, but not today, even though it can do everything else I want it to do. It has just 8GB of memory. I don't remember what the GPU is, but it was supposed to have been top of the line 13 years ago. I actually went to a computer store, got a shopping cart, and asked an employee to walk around the store with me to pick the best of everything, from the case and power supply to the CPU, GPU, hard drives, and Blu-ray writers. I suppose, when this one breaks badly, I'll have to build another and switch to Linux.
I have over 100 TB of storage, with hundreds of thousands of books, magazines, and newspapers, thousands of short videos, a ton of TV shows and movies, and over a million music files. On top of having a local AI search all that and index it, I would also eventually want to access it all remotely with a cellphone and do it securely.
I want a lot, and I'm in my 70s, so I also want it soon.
Whatever you figure out would put me far ahead of where I am now. I took computer classes many years ago, and before that, around 1975, I started my self education, buying my first computer in 1982. I have learned SQL and PHP on my own just enough to program stuff on my own websites. I have learned some about AI and the LLMs, but I have a hundred other projects going at the same time, including writing a book for a rich guy in another state and then continuing writing books for my own account to sell online.
First off, thanks for this comment. Its the kind of discussion I come to this site for.
Its typical Nino strategy. He always plays devil's advocate, on behalf of this more black pilled listeners, and it always works. Right now, we have too many patriots who have let the pendulum swing too far to the right, so to speak, when it comes to anything "digital", esp after the Plandemic days. So, this kind of devil's advocacy makes them feel like they are being heard, and their big fears are being addressed.
He did the same thing during "46" years, always complaining to the point of sounding blackpilled himself, while letting himself be talked out of that despondency by the various people he interviewed.
There is also the aspect where he cuts of juicy bits so he can discuss them on his payed channel, often using youtube censorship as excuse. I think in this video there wasn't a whole lot of that just one or two instances.
I hate using the word AI as much as possible, so I will use LLM (large language models) because for firstly thats what they are - simple language models, and secondly, it reminds us that there is no "intelligence" and they are not "beings" but just tools.
So the utility of LLMs depends vastly on how you use them. They canbe extremely powerful if wielded correctly, but extremely disappointing if wielded wrongly.
Imo, wielding LLMs correctly requires us to decide which part of the hard work we want to unload on them, and which part of the work- esp thinking aspects - we want to do ourselves.
For instance, not withstanding the fact that it offered your Uncle's name, I would argue that LLMs can be very powerful in scraping genealogy data, compiling them, deducing connections through large number of clues that might be hard to manually analyse, and then build a system that would provide the answers, rather than using directly to find answers.
LLMs are also great at obtaining a list of very relevant source material for nay research, summarize them, prioritize them, while the human does the actual research.
I have been a coder for 35 years, and my entire perspective of software development has completely changed in the past 3 weeks after GPT-5 was released. I can now build complex software in minutes and I haven't written a single line of code in the past 3 weeks while i have build 20+ apps for myself, and even though I am a hardcore coder, I am enjoying this experience.
And the growth of LLMs is exponentially faster than any other technology. The day I first accessed the internet in early 90s, to the day it became a somewhat useful tool for me, was almost 4 years (even though I used to be engrossed in the early days of internet daily). And almost 10 years before it became a tool that non technical people can use.
Compare that to the day Meta released their first LLM - April of this year. And in 6 months, it has completely changed the way I have been doing a core part of my life for 35 years.
This is one thing I want as well. They have something called RAGs right now and I set one up, but they dont work as well right now. But the quality of LLMs, and esp the reasoning LLMs is growing so fast that I think in the next 2 months I will be able to build a system like this that would work better than a human. If I do it, I will definitely ping you.
This is my personal passion as well. For the longest time (since April that is) I have been using local LLMs running on my own Nvidia to do my work. But unfortunately the computing power is too little and the quality of these LLMs is very limited, when you compare it to GPT-4 or GPT-5, for instance.
So I have shelved this passion for now, and embraced GPT-5. But I know that as the math behind LLMs become advanced, and the GPU hardware tech keeps advancing, pretty soon we will be able to run GPT-4/GPT-5 level LLMs locally.
I think that would be really necessary for companies working on sensitive information to be able to start using LLMs well.
Depending on how you define "real money" to be, if you use an objective definition that money is something that has inherent value, then only a quantum of energy or a quantum of computational power will qualify as "real money". These two have the power to materially transform things, and hence have inherent value.
But if we define money to be the something that people are willing to exchange for something of real value - still a subjective definition - gold and silver are by far the closest to real money
The problem is not even that. The real problem is that money has stopped being "thing of value" and has become "debt".
This is the beauty of "Zero Knowldge Proof" actually. You dont need an external computer server to carry out the transaction. If I were to want to verify your age, firstly, you will only exchange data with me. Neither of us will need to exchange data with any other server.
You will have a encrypted certificate on your device given to you by the authority that issued your ID. I will have an encrypted key of the same authority on my device.
I will use my key to validate a piece of data generated by your device using your certificate and thats all. There is absolutely no necessity to talk to any external computer.
Of course, some people may choose to store their private certificates on a cloud server, but thats like people leaving their bitcoins on the exchanges. Its a stupid thing to do, but people are free to be stupid.
This is allthe more reason why we need to put in a trustless decentralized system in place, so that future generation does not need to worry about it.
But making all this work ina. user friendly way while still keeping it secure is a challenge that comes in at implementation phase. What I am really concerned is for the digital warriors to always watch like hawks on any implementation of these digital ID systems, to ensure they are zero proof, not centralized and protects privacy at source - not sending any private data to any servers, period.
Thanks for the reply.
I refer to "real money" as what the US Constitution says. Congress was supposed to "coin money," but they gave that job to the Fed, and then started doing a bunch of other stuff that's unconstitutional. Also, gold has always been money for thousands of years. It's physical and fungible. Anyone can understand it. I'm still on the fence about blockchain money. That's another subject I need to research, if I just had more hours in the day.
I will be glad when I can have a computer system at home that has the horsepower, memory, speed, and storage to do what I want. I built my current computer 13 years ago. It was state of the art then, but not today, even though it can do everything else I want it to do. It has just 8GB of memory. I don't remember what the GPU is, but it was supposed to have been top of the line 13 years ago. I actually went to a computer store, got a shopping cart, and asked an employee to walk around the store with me to pick the best of everything, from the case and power supply to the CPU, GPU, hard drives, and Blu-ray writers. I suppose, when this one breaks badly, I'll have to build another and switch to Linux.
I have over 100 TB of storage, with hundreds of thousands of books, magazines, and newspapers, thousands of short videos, a ton of TV shows and movies, and over a million music files. On top of having a local AI search all that and index it, I would also eventually want to access it all remotely with a cellphone and do it securely.
I want a lot, and I'm in my 70s, so I also want it soon.
Whatever you figure out would put me far ahead of where I am now. I took computer classes many years ago, and before that, around 1975, I started my self education, buying my first computer in 1982. I have learned SQL and PHP on my own just enough to program stuff on my own websites. I have learned some about AI and the LLMs, but I have a hundred other projects going at the same time, including writing a book for a rich guy in another state and then continuing writing books for my own account to sell online.