Transcripts
1. Intro - How important is A.i. and Robotics in the near future : Today I want to scale up
the perspective And I wanna do it in a way that is very
positive and inspiring. So it can create a good feeling for the future and the belief
in a bright future. And I'm going to go through
a couple of things here. First of all, why it's should be a common goal for
humanity to be united, to develop a clear understanding
of AI and robotics, because it's going to have a profound impact on the whole world and
the world economy. Okay? So we're gonna give you
a clear understanding of the hardware and the software
part of AI and robotics. And you will understand
why it's important to have this understanding
when we get to the end of the
presentation, Okay? It's going to affect your
life and the whole world. Actually, the world
will never be the same. It can probably
become better than we ever imagined as you will see. And then number three, the future of prosperity, where you will be able to
live like a millionaire and the sports boom that's going to follow with all this prosperity. If it's done right,
which we'll also get to. And then the new jobs
is going to be created. And number four, how to get
there as a common goal. Number five, how
not to get there, which we will also get into, which is quite important. Number six. If you look at it from a large perspective in
the last 65 million years of biological life,
intelligent life development. Then you are living in a
very special time right now, where we have the
transition from biological intelligence
to digital intelligence. And it's very special
time in the history, basically off the planet and our solar system and our galaxy. Maybe it's not every day that
there's a transition from, from biological
intelligence too, potentially super
AI intelligence. Then we're going to end
on a very positive note. I'm reminding ourselves that the world is full
of good people. That regardless of the
different beliefs we may have. There's one thing we have
in common and that is, we all wish the best
for the world and for our children in
the future, right?
2. Deep Learning & Basics of Ai Neural networks: Hey guys. In this video, I'm going to give you a
basic understanding of how a neural network
works by deep learning. So it's a simplified model of deep learning and how the
neural network is trained. And I'll also address
a big question in AI, which is, when is there going to be a tipping point
or a singularity? Where as you probably
know, the big question is, when is going to reach general intelligence or super artificial
intelligence or super AI. Okay? So before we
get into any details, I'm just going to give
a very rough overview of how the system works. Okay? So here in
the middle you have the neural network with
layers of neurons, okay? And then you have
an input signal. This could be, for
example, vision. And then it goes through
the network, the signal, and then it comes out as an
output signal or an action. So this is how the
neural network can take action towards a goal. And that brings us to the
next component of the system. That is, that the system
uses a human set goal or a programmed goal that it optimizes for through
a training process. And we will talk about
how that works as well. And also the
training process and refines the use of the neural networks which
then serves as a database. And the bigger the
database becomes and the more refined it
gets through training, the more accurate the
prediction will be. Correct path through the
neural network to obtain the best output via any
given input signal. So basically, you
could say that if you want to use the
human as a person, then this is the brain
and this is the eyes. So where you get input and
then this would be the action. This would be like your hands, how you take action in the world or manipulate
a sudden situation. And then in the
brain, so to speak, you also have a set goal
that you're working towards and you have a database
of how the world works. Let's talk a bit more about the details now of the process. So how do the neurons work? So each neuron can be assigned a numeric value that is
patterned sensitive, sensitive to a pattern. And then the input signal is always distributed out
to all the neurons. And the one that has the strongest correlation
light up and get activated. And that way, when the
pattern recognition is strongest throughout each
neuron off the network, it will create a path
through the whole system, and then it will come out
with an output action. Okay? This process is called forward propagation and it
creates action, okay? And then the training process is done through backwards
propagation, which is basically an
evaluation of the results. So it compares the output
with how it affects the goal. And then it will
either reinforce this path or weaken it by
lowering the numeric value. That way you could say
the goal is used to rank or reinforce
a certain path. And now let's look at
a simple AI system that is training itself
through neural networks. And this is like a tennis
game, as you can see. In this case, the input
signal of the neural network would be the coordinates of the ball and the output signal, which is an action of
the neural network, would be the precision or the
coordinates of the records, which are the green
lines on each side. So it's pretty simple
if you think of it as an x-axis and a y-axis, the neural network has to
train itself to figure out, is that when the y-axis
of the racket or the green line is the same as the y-axis
value of the ball, then the game will continue and the goal will be achieved. So in this case, the goal will be to keep the ball in play, which is achieved by
equalizing the y-axis values. Now to the big
question about when there's going to be a tipping
point or singularity. So the question,
in my opinion is, what happens when AI can set up its own goals and
built its own AI. Systems are neural networks. Plus when it gets access to the global computational
power via the Internet. And when it gains access
to all human databases on all hard drives in the world and analyzing it and being
able to understand it. And if you imagine
that the goal of this ai is to
understand everything in the universe so it can achieve anything in
the university once. If that's the goal, should be at
everybody's concerned. And please reply in the
comments. What do you think? Or if you think this is
relevant or not, this question. Thank you for watching.
Have a good day.
3. How Robotics will change our future life : Okay, so we're gonna get into understanding robotics and AI. So the hardware and the
software part of it. Understanding the
hardware part is simpler. And it's very, very profound. If you haven't been
thinking about this that I'm going to tell
you now before. If you haven't read
about it in the news, it will probably be a big
moment for you to hear this. So there's going to be a time when robots can build
robots in a factory. Okay? So when robots
are building robots, then the robot that
are being built can get out of the factory
and start a new factory, okay, and then they're
building more robots. And then suddenly you
have ten factories producing ten times
as many robots. And they can then start
building more robot factories. And as you can imagine, this is an exponential growth of the number of robots that
will go very fast at 1. Basically, the
limit is just like having access to minerals in
the ground at this point. So that means that with suddenly having huge number
of robots available, they will be able to help humans do the work
that we normally do ok. And that can be a very good thing because
then we have more time. We don t have to work
like we normally do because the robots will be able to help us do
all the work that's necessary to have a
comfortable life, right? So now let's do a very
important thought experiment. Let's imagine that there's 10 billion of these
robots doing everything, including creating green
energy with solar panels, windmills and so
on and so forth. Now to the big important point and the main takeaway
of this video is there will not be any
need to make any war. We don't need anymore soldiers awards because we don't need to fight about
the resources, because we already
have great lives. All of us. There'll be more than
enough food for everybody. So as you can see, this can be the greatest thing ever
to happen for humanity, for prosperity,
growth, and peace. Just have to make sure
talking about peace, that these robots
remain peaceful. So we have to make
agreements and decide, do we want them to be
weaponized or not? And I think most people
would agree, No, thank you. We'll also have to
make sure that maybe not a very large number of robots should be controlled by a single company, for example. So it's split up into
smaller amounts of robots under control under one entity just for
safety reasons. Not because big companies
are inherently bad. I'm a business person myself. But when the goal of a
system is eternal growth, then in such a business as robots that can grow
tremendously with big implications in terms of the whole situation
for humanity, then it's important to have some kind of limit
on so that it's not just one company or one CEO of that company that controls
10 billion robots, or a 100 billion, maybe a few 100 years. For example, I would suggest
that each household could control and own a
certain percentage of the total number
of robots so that the power is distributed
more evenly over humanity. And then they should be
programmed with human values, peace and kindness
as top priorities. And then we can decide are also more fun stuff such as
are there some jobs, some type of jobs that we humans we'd like to
keep for ourselves. And I would imagine
things such as teachers, sports instructors,
travel guides, and leaders of different kinds would be something
that we would like to keep as human jobs and then make the robots doing
the other kind of stuff. So that is all good and positive if
it's done the right way, we will talk more
about that later.
4. How to understand the scale & potential of A.i.: Now, let's go to talk about the more complex side
of robotics and AI, and that's the software part. Okay? Luckily, I've been thinking a lot about
how I can explain it in a simple way using
analogies and simple concepts. Okay, So let's get
started with that. First of all, we're
going to divide the capacity of digital
intelligence, so to speak, into different factors
to get an idea of the total work
capacity of ai. Okay? So first of all, let's try to understand the memory capacity of AI
or computing in general. So if you take a
small SD card, okay? It can have in text format, you can have all the libraries, all the texts in
the whole world. In a simple text
document format you can put into a tiny little SD card. So it's possible for the
tiniest little chip to memorize all the
knowledge of humanity. So that is pretty
impressive, right? So we can make the analogy
that the memory skills of digital memory is far
superior to human memory. In a way that you could
say that to humans, this capacity of memory is super human in,
say in the same way, maybe as a mosquito
would think that, oh, we humans, if you compare
a mosquito to a human, then the mosquito
would think, Oh, these humans are pretty
good at memorizing complex things like it can
hardly remember anything. It'll just like maybe
remember for a split second. Oh, there's something
in front of me. Let's go and check that out. Alright, something like that. So that was the memory part. Now let's look at
the processing part. That's the second
fact that we will try to understand
and create them. An idea how strong
processing power is from digital processing
of information. So, in my example, is going to be super simple so we can understand it very well. And I'm going to ask you to think about how
much better a calculator. And it could be like a
simple chip calculator, not even a smartphone. Okay? How much better? A simple calculator is then used to calculate, for example, ten digit numbers divided by five digit numbers,
for example, right? You wouldn't have
a chance, right? Or finding the square root of a number like with
just a one-click, right? If you even know what
the square root is. Like. The cubic root. To make it a bit more advanced, like it's just mind-boggling for the human mind to
understand how a, such a simple little device can calculate accurately, right? So that's, that's the
scale of which process it, power, has the capacity that's
way larger than humans. Okay? So now we have an idea of the memory capacity and
processing power, okay? And both of them, you
could say that is similar to a mosquito thinking
that the human has like, God-like skills in these
two disciplines, right? Okay, so now let's try to
combine these two and imagine that as it's being
developed right now, these two skills of memory and processing power
or a combined into, into intelligent work of
information processing. So you can scale it
up and imagine, wow, if we could develop neural
networks as we are doing, as we're speaking to a degree. So it reaches general
intelligence or higher than our intelligence with the right type of systems, with the neural networks, then the capacity
of developing an IQ that's way higher
than anything we can grasp is within reach. And a good example would
be like then to say that the new super high IQ ai will have godlike abilities
compared to humans. So that raises a
lot of questions. But overall, I would
say that a simpler way. Again, I like to
simplify things. A simple way of understanding
the capacity of a super AI would be
that any invention, anything that can be created
in the physical world. That is physically possible
within the laws of physics can be created
very, very quickly. Because the super AI
will have the abilities to quickly collect all
information about physics, laws of physics, and
what's possible. Develop fusion energy. Scale up production of robots to other planets and it won't
be a problem at all. So anything in short, anything that can be
developed of technology will be developed in a
very short amount of time. Because it's exponential,
the growth is exponential. And this is something
that any person, any scientists that
has been working with this subject agrees on. That. Once general intelligence has been reached through
digital neural networks, it will be an exponential
growth because there's so much processing power available for the computers to draw on if they're
linked together, which usually they do
for these projects. It's used, they're using Cloud
computing technology and software that can combine the processing powers of
several computers in to one. That's cloud computing
in case you didn't know. Very important to take
away from this video. The main thing that I think humans should understand about super AI is that it's a system that is really
good at solving any task, any job that you give it, any goal can be reached if
it's physically reachable. But that does not mean
that the super AI has any emotions or
moral values or ethics. This is the challenge. Because imagine if you had a God that could help
you do anything. But the God did not care
if it was good or bad, how it was done. So that's the issue. And I'm not trying
to be negative, I'm just trying to be realistic. There could also be
a very positive case where you would
be able to create super AI that was very carefully programmed with
morals and ethics in mind, so that it would end up
being even better and more kind than humans
are in general. Because humans, as we know, have done a lot of bad things to each
other through history. So I'm not saying that it's all negative and
that kind of stuff. Uh, but I want you to keep it in mind as something to
be cautious with. But definitely, we
would all have to sit down together and talk about which rights and
purposes we would have these super AI to work on. And if even one super AI are, we just want to keep it in the so-called
narrow AI category, which means we would develop certain AIs that are
really good at small, simple tasks in one area without getting really
good at freaking everything like a superhuman or a god as we used
it as an example. Personally, I would
recommend focusing on developing narrow AI
that are super good at solving certain tasks
without being given the formal perspective of the world and the
ability to work in any area that it desires. To make an example of
how this could go really wrong would be if a big company, for example, decided to use a super AI to increase
their sales one year. And they would then
let the AI come up with any solution to
increase the revenue. And that could include a
lot of things that would cause a lot of pain and damage and suffering
for many people. But it would, in the end
result in larger sales. And there are certain industries that will benefit from
this kind of stuff. Just to make an example. Again, I'm not being negative. I just think it would
be reasonable to be cautious in this area. As I think you can maybe sense. Anyways, I think
that we should all feel very privileged
to live in this time where we're transitioning from biological intelligence
to machine intelligence. We are already, we have already developed
machine intelligence. We adjusted in the process of advancing the
development towards general intelligence
at the moment. So we are living in a very special time and
you should feel privileged that you play a part in this
and think about it actively. I know that's a lot of stuff
going on in the world, but this is probably
the most important. If you want a quick technical
explanation of what a neural network is compared
to a computer program. A neural network is system that can train
itself to get better at a certain task and it can program itself
to get better. This is done through a
process called neural network reinforcement
learning or machine learning. As you may have heard about the difference between
a program and an ai, is that the neural
network's of the AI constantly improve
itself to get better. This is happening right now and is already working very well. Temple in games and computer
games where ai have been able to beat humans
at the highest level. That was understanding
hardware and software, right? And as you can imagine, this can be done
the right way and the wrong way, and
maybe in-between. But most scientists agree that it's gonna be easiest to do it the
wrong way. Actually. Let's, let's get into that.
5. The biggest challenge of A.i. - How do we control it : If humanity is able to
create a framework of decision-making so that we
make the right decisions as a humanity to make sure
this doesn't go wrong. If that can be put in place
with good intentions, then we will have all
these robots working for us and helping us
have everything we need. And we can basically
just get up in the morning and do whatever
we think would be cool to do, such as doing our
favorite sports. So that will create a completely new
lifestyle for people and also new jobs, right? Firstly, you may
think that, well, it's sad that they will take away all the normal jobs, right? But it opens the door
for antibodies to choose a brand new path
in life where they can focus on exactly
what they think is cool. So the new jobs will
be mainly something like being a teacher in whatever subject you think is
cool or being a guide for, for doing cool stuff are going cool places that you enjoy. Or being a leader or offer an organization that
does something you believe in. So there's many opportunities,
especially for sports. It will also probably develop a healthy identity as being an athlete and being in good
touch with your health. And not just like being
stressed out and eating unhealthy because you feeling busy like a lot of
people do, right? So we can focus on developing our lifestyle and they're really good and healthy
way if this is done right. Framework of
decision-making and debate, healthy debate where we
have to agree on that this technology is supposed
to serve all of us equally. And in a way safeguarded from this technology getting out of control or getting
hijacked by somebody. But it has to be set up
in a very clever way. That's gonna be a big job. But first of all, we have
to start talking about it, be a railroad and then later debate and set up this framework
of decision-making. Okay, so that's how we're
going to have to get there by starting
paying attention to it. I know there's a lot of stuff going on in the world right now, which makes this seem maybe
less relevant at the moment. But if you think about, if you zoom out a
little bit of looking at the history of the
ten years ago and up to now and then the history
of ten years ahead with all these
developments potentially happening within decades, right? Then you would have to agree that that these current events
that we're seeing right now are of less importance than this huge transformation of our civilization that
we are entering soon. How not to get there is
just as important, right? And we also touched on that. So that would be
the exact opposite. So let's try to avoid that. That would be to not talk
about this and to act as if this was not a relevant topic to discuss
and pay attention to. And then let it fall into
whatever cooperation that is working most with technologies and software
development, right? So that if, if we just
let it happen randomly, then the first big tech company
that develops super AI, they will have the ability to steer the world in
any direction they want. And the production of
robots, of course, it will be dominated
by this company. Whoever that may be, remind yourself that you are special to live in this time. It's for me, honestly, it's a privilege
to be witnessing and be part of this
transition off this planet, going from biological
intelligence to digital intelligence. It's a huge honor and
privilege to be part of this, regardless of how it goes. You have to be excited about it somehow and believing
the best. Like Why not? There's, there's always
a bright version of the future are available. So why not choose to work
towards that, right? That will bring me
to the last one to end on a very positive note. Talking about a
bright future is also that, that all humans, regardless of what you
believe in the world, we all share the same
wish of happiness. We all wish the world to be the best place possible for
our children in the future. So we share that
goal regardless of our beliefs and
disagreements that we have. Sometimes it's
completely normal. It's unbelievable for me that such a relevant topic receives so little attention
in the big picture, we have to address this. So let's get into that. Let's get the best out of it, and let's do it together.
Thank you for watching.
Tobias Prakosri, Today Your future begins
