What would it take to demystify chaos and the order of the universe?
Let’s get down to this delicate question, discussing Artificial Super Intelligence.
As far as I understand, there are quite different opinions mentioned lately by the main protagonists in the field. The huge progression in the field of Artificial Intelligence is shaking up the developers as they digitalization to see possibilities and limitations of this new technology.
The hype in the field of Artificial Intelligence is reasonable as there are lots of possibilities for a possible last step in digitalization.
Questions about Artificial Super Intelligence are of a philosophic kind and not reality in any way!
Read my conclusions for a deeper insight …
Artificial Intelligence is no Artificial Intelligence!
As far as of current development we are at a stage of Deep Learning in the evolution in the field of Machine Learning. Computers are getting some form of intelligence, at least some of the developers and researchers think they do.
Are they really intelligent and are they also artificial?
I don’t think so and I have good reasons!
To get into this argument, let’s first look at the basis of Machine Learning.
Machine learning works by optimizing statistics.
When a Machine Learning program starts the famous task of analyzing an image, it, first of all, is fed by data. This so-called training data gets processed by the various algorithms. These algorithms are trying and guessing at the beginning to get correct results concerning the tested subject, but also save their guesses and correctness of results.
When being fed with millions of similar objects to analyze, they start adjusting their guesses as a function of their saved results from prior problems they faced and optimize the outcome.
They are approaching reality and after being trained they are able to optimize their guessing.
First of all, it needs a lot of computer power to solve all the mathematical statistics faced to figure out the guesses and our computers are now fast enough to handle huge amounts of calculations through GPU’s (Graphical Processor Unit) and maybe even through quant computers, becoming available now for some of the biggest players.
The effect, the teams of Machine Learning at the various labs are experiencing is a fast adjustment of the algorithms when being trained by the huge amount of data available in our connected world.
Even the creators got shocked by the fast learning capabilities of these algorithms by now, reaching an accuracy of predictions higher than humans in different fields like image recognition and many more fields.
The accuracy of predictions is seen as intelligence. Is it intelligence?
Subscribe to receive my constant free updates about the progress and opportunities of the online markets.
Statistics and their limitations
What is statistics?
I think nobody will argue with me if I argue, that it all comes down to the Gaussian Distribution of errors.
When measuring experiments it worked all fine for us during the evolution of science when using the normal behavior of the field of study.
When looking at simple physical experiments like a falling stone, the normalization is highly accurate, proves the equation of F=mg and is very useful in all cases we need to build our mechanical machines.
Mathematics compared to Physics
Physics comes from the ancient Greek language and means “Knowledge of Nature”. In Physics we are trying to understand, how the world works and we have been very successful, using Math to describe the problems faced.
Math itself has nothing to do with our world and nature. Math is a completely logical construction, figured out by mankind in thousands of years, using different approaches to describe the theory. Besides the simple evolution of our casual need to use functions in our lives, there is the pure theoretical base of Math, where we define a detailed basic concept. starting with simple numbers and functions.
Relevant to our lives are equations like 1+1=2, 1+2=3, …
We are using this for a very long time and it simply works for our logically oriented brain, maybe through instinct for survival.
More complex situations
We found out, that our environment is a lot more complex. Mixing and joining equations led to solutions of more complex problems and human intelligence searched for a deeper understanding of the environment.
Physics have always been a main driver for the development of Math. The rules we figured out for nature are not obvious and it requires profound observations to detect how nature works.
When searching rules, attached to a physical problem it turns out, that most problems are not isolated and can be broken down into various logically connected problems. Isolation of the various problems has proven to be very effective to find the underlying rules and set them up in Math.
I find great value in applying this concept to many real life problems. Separating problems and levels of observation result to be very effective for solving any complex problem.
In simple problems, these connected problems do not disturb the logical rules, like the equation of the falling stone.
(Looking at the issue from a modern overall perspective we had to expand the theory to include the general concept of gravity, where the Sun, the planets and the entire Universe interferes with both, the stone and earth, causing an error, irrelevant for most casual problems).
When measuring Physical experiments, scientists found out quite fast, that their experiments produced errors. All measurements couldn’t be precise due to mechanical possibilities faced, like time measurement, length, etc.
There have been many approaches for standardization of elementary parts of measurement, like the Kilogramm, the Second, etc. but it turned out, that we had to dig deeper down to atoms to find sustainable solutions.
It turns out, we found quite elementary definitions to use Physics to build all kinds of machines and even fly to the moon. (We tried many times and even when we figured it out, some of the missions have not been successful, but normally we did it … some space shuttles crashed as part of the error)
Isolation of the problems was quite successful and through Gaussian normalization, we found a very efficient way to keep errors in control.
We built our empire on it!
The point I try to make here is, that Physics and the Knowledge of Nature have errors, but for most equations, they just don’t matter.
Math has no errors at all!!
Math as a science is a construction of logic, closed in itself and pure!
Errors are modeled inside the construction Math by expanding the field of Analysis and Statistics. Math defines, what to conclude from experimental errors.
As by common sense we know “Division by 0 is not allowed and invalid”, but in the construct of Math, it is defined as “undefined” as the division function runs into infinity through approaching small values near 0.
Computers and Errors
Besides the errors resulting in nature and measured in physics as discussed above, we included errors to reduce complexity and make the problem easier!
In Math 1/3 = 0.33333… (never ending), but when we use it, we might calculate with 0.34 or 0.33334. Results in projections are never as exact as 1/3.
This brings me to the first point:
Computers implement mathematical functions, but they don’t work with mathematical abstractions!
They always work with rounded values, defined by the size of the dedicated storage capacity as described by a float, int, long etc.
To process equations, you need processor speed and you need storage to store it efficiently by size and speed.
The progress made since the beginning of computing made it possible to calculate and visualize things we would not have explored without the machines we invented to support our capacity for calculations.
But these discussed errors consist in the basis of computation, even if we have them quite well under control.
Chaos, the mystical Knowledge of Nature
The progress in measurement and computation led to a discovery of the special field of Chaos and Fractals. Effects have been detected same in Physics as in Math.
These Effects had been seen as normal errors in experiments and have just been accepted until Computers had speed up to provide enough data to visually identify the phenomena and cause the identification of the discovered problem.
Chaos in Physics is not, what we all know as similar to a mess! Chaos describes specific physical effects. These effects have in common, to be nonlinear and not to be predictable, how to act in the future.
If you start a pendulum it will run for some time, slowing down linearly, caused by friction. But it doesn’t seem to be likely, that it starts swinging like crazy again after some time. The pendulum is predictable!
There are physical processes we can’t isolate, extract or encapsulate!
There is no solution to find out exactly, where water particles are floating when a river is running down the valley.
We can’t predict how smoke is floating in the air!
Math as a science worked on these problems and they developed complex concepts to approximate the real circumstances but the unsolvable error stays there.
These effects are known now as “Butterfly Effect”, where a butterfly in Peru could cause a tornado in India.
It results, chaotic processes are no strangers, we can find them everywhere!!
As we found out, chaotic processes don’t just exist in real physical experiments but also in special mathematical functions, mainly as an effect of computation error rounding.
Conclusion: Slight variation in the starting values can result in completely different results after some time.
We can observe this problem in wheater channels, where they nowadays show different prediction models, diverging heavily over the time of a week.
Mathematicians worked on strange models in their field for a long time and discovered some geometrical expressions without solutions in classical Math and Statistics, expressed by fractals.
Fractals are caused by geometrical iterations but also follow other mathematical rules and functions.
Experiments with visualizations of chaotic mathematical iterations resulted in Fractal Shapes, being of complete order and very similar to creations of nature.
Physics and the calculation of the universe
For a long time physicists argued, they would be able to calculate the Knowledge of Nature if they would just have all the needed information to feed the formulas.
They have been taught wrong! They can not calculate the Knowledge of Nature!
In many fields, they have made great progress. Applied Math in Theoretical Physics led to conclusions about particles, that have to exist but couldn’t be measured.
Great minds like Einstein and others developed theories about relativity and quantum mechanics, explaining the universe more and more profound, but this is a further chapter …
The more we know the less we know!
Religion and the end of imagination
Whenever we couldn’t solve a problem in history we found an easy solution in delegating the problem to God.
If there are no solutions presented to our imagination, there has to be a force running the show behind the scenes.
I am raised in a very open minded Christian family but I don’t believe in God in a way, any religious institution thinks about it. Be it God, Allah, Buddha, Olofi or whatever you name it, God just doesn’t work out for me, but I do believe in a naturally given order or strength, keeping it all together. Not just on a physical part but just entirely.
I can’t find any discrepancies with Physics and Math but I also do not exclude what I can’t see or measure.
The conclusion I take out of all this philosophical thinking is just living the best way I can and experience the unknown without end of imagination.
The rise of computer intelligence
Computers are getting more intelligent in specific fields.
We follow the same approach as when building our knowledge about Math and Physics, progressing in various fields and causing real revolutions.
Scientists are arguing about creating Artificial Super Intelligence, just amazed by creating a new God to delegate.
It is not that easy my friends!
Let’s suppose you could generate an Artificial Super Intelligence and according to my understanding, you would do this by an eternal algorithm, able to learn anything and just figure it out.
Let me ask you some questions now!
Figure out what?
How do you plan to eliminate errors and resulting chaotic behavior of your machine?
How do you plan to handle the chaotic behavior of nature, your surrounding?
How would you know when to start the learning of the machine?
Maybe somebody should prove this, but I wouldn’t be surprised if these basic chaotic principles would be the reason for the creativity of living creatures.
Just as a beautiful example: When observing how our eyes are moving when looking around, exploring the world, the underlying process has been proved chaotic.
Philosophically spoken: Nirvana can not be built, Nirvana exists
The argument of consciousness
I would like to include the discussion about consciousness in this article as it pops up a lot in the different AI lab as the main argument against Artificial Super Intelligence.
Even if the community of AI scientists and researchers seems quite united and willing to democratize (whatever it means) the AI technology, there are some discrepancies between young and old. the young generations fascinated about all the stuff that is becoming possible.
The young generations are fascinated about all the stuff that is becoming possible.
Now Mars is in reachable distance, sci-fi is becoming reality … it’s a little spooky and there might be higher forces involved … but there is also the devil around and antique cultures …
The older, including my generation (pre millennials), have seen the rise of the industry and the computer. We have seen the good, the bad and the ugly and we fight for our beliefs.
Amongst the older generation of AI scientist, there is an argument going on for more than 30 years about the possibility of creating a consciousness of an AI and they are coming up with the ever old argumentation about consciousness.
The argument is absolutely correct in my point of view but I just reduce it to the question:
How should there be a sense of doing something explicit for any so called AI if there is no specific goal pre-programmed?
We are talking about the sense of life here, guys!
There are other fields of science we need to treat apart, especially Quantum Mechanics and Biology …
Philosophy in action
We, humans, built incredibly complex systems. They are fragile and they can run into deep problems, but until now, modern civilization has not faced a collapse.
In my life since 1978, I have seen some quite dangerous moments and heard about even more critical moments in the past.
Isn’t it incredible, how we can live on earth in such a complex community, making use of all things and just live until we die?
When I was working in a private bank in Switzerland in 2007 as a mgmt. assistant, there was a time when I thought some part of our real economy could collapse. What if there was no fuel to bring food from the country side to the big cities if electricity would fail, payments could not be handled and the stock market couldn’t get out of his chaotic movement, destroying wealth but also destroying perspectives.
I was astonished by how the problem was solved in a worldwide common sense and did not collapse.
Will the world collapse through AI?
I don’t think so, but we should take care of how we treat this technology.
The world agreed to not attempt to clone humans. The same we should think through in detail for AI!
Bias and training data
Let’s step back from the Nirvana solution, as in my point of view, there is no sense to discuss this issue any further.
But there are arising problems resulting from some of the thoughts discussed above!
Let me draw you a process of the danger and it starts again in statistics. You can collect all kinds of data and do complex analysis on the base of the data.
If you ask non-sense, non-sense is being returned!
From Chaos, we know, that a minor change in the starting environment can result in a completely different state after some time. If we integrate this idea into the multiplying effect of layers of deep learning networks we might have to ask the question how we confirm constance in the behavior of a complex AI, like they are being built currently.
The question about giving the machine a certain sense of life also results in a limited knowledge, as it is always based on the most average thought.
If we think about it, we collect the data we get, but the extreme case of somebody living in the forests without any connection to the internet and his experience are completely irrelevant for the algorithms.
US elections might be an example, where data about conservative people without a connection to new media have been left behind in the predictions. (There have been many other influences …)
How many times did people out of nowhere create extraordinary stuff?
By applying certain data to the AI algorithm while he learns is shaping his idea of the world. Different cultures have different common sense and in some states, the new AI services might decide about life and death of discriminated individuals.
I think and believe that it is a right of biological creatures to have a brain of a limited size. I think something most essential to our brain is the capacity to forget and optimize at the same time.
Our brain has the perfect size, perfect power handling, perfect input devices and our body to play with and learning in different cultures enriches our experience of life.
The new technology we now call Artificial Intelligence in my point of view has nothing to do with AI and any expression similar to us humans.
The 4th tech evolution systems we are creating are extending our intelligence and challenging our ability to interpret.
In many fields, this Extended Intelligence will help us solve a great variety of problems, besides creating new ones.
I think we should not see this as intelligence at all! In fact, it is just applied analysis systems!
These applied analysis systems will change the world as we know it now, but the world always changes!!
Even so, I see it as crucial to becoming familiar with the new tech wave and transition sooner than later!
In a global digital world, I think it is crucial to learn how to expose an idea on the internet. It will be challenging so I recommend the approach of The Six Figure Mentors to start the journey into a digital world.
As discussed, in my point of view, any interpretation of Artificial Intelligence in the current fields of Machine Learning and Deep Learning has to be taxed as artificial bullshit and in my point of view, this is not a game, guys!
We should stop developing systems with general purposes and get down on the specialized problems in detail and without excluding the errors, hunting them down.
Let AlphaGo play Go, don’t teach him chess and better create AlphaChess. Got the message?
I very much like the idea of what the game did to the community in opening up space for new concepts in an exchange with the machine.
This is what is happening now. We can attack REAL WORLD PROBLEMS and SOLVE them through the technical evolution.
Let’s use it for good!