What is Real AI? Is it AI or lack of intelligence? We’ve all been talking about artificial intelligence lately. Whichever website we enter, which article we read, which television channel we open, we see an artificial intelligence. We are faced with advertisements every day claiming that even electrical household appliances work with the support of artificial intelligence.
When we sweep our house with a vacuum cleaner allegedly equipped with artificial intelligence systems, does it suddenly tell us, “I’m tired, I’m not going to sweep anymore, or we’d better clean this place like this”? Of course it doesn’t, because an automation that lacks the ability to reason cannot be called artificial intelligence. At best, it can be expressed in such a concept as artificial “incomplete” intelligence. Every automated device is not AI!
What is Real AI | Is it AI or lack of intelligence?
So we are experiencing an artificial intelligence inflation in which even simple automation algorithms are called artificial intelligence. Sure, automation and learning-based algorithms are making our lives easier as they evolve, but it’s hard to attribute “intelligence” to them. Artificial intelligence needs to consciously learn on its own without external guidance and evaluate, reason, question and make decisions based on what it has learned.
Take a look at the following historical associations in order to purify the concept of artificial intelligence, which is about to undergo an axis shift in social perception, and the research and development activities in this field from all kinds of industrial utilitarianism and direct it to its real goal and to participate in the social enlightenment on this issue.
What is Real AI? Every robotic system means AI?
Discovering: Intelligence is a collective phenomenon made up of numerous components. Let’s try to relate this phenomenon to “exploring.” Of course, we are talking about “exploring” in a conscious, systematic and questioning way, and by using the opportunities we have in a strategic and optimized way. Albert Einstein, for example, was one of the smartest people ever, discovering that space and time are inextricably linked. More importantly, he achieved this with the help of systematic thought experiments and mathematics. Again, J. C. Maxwell was another genius, he discovered electromagnetic fields.
Marie Curie, known for her discoveries in the field of radioactivity, and Richard P. Feynman, who invented the Feynman diagrams that made it easier for us to at least “try to understand” quantum mechanics, were undoubtedly very intelligent people. We can easily reproduce the examples.
The common denominator in all these examples is this: pure, true “intelligence” that learns from available data, that consciously and systematically combines what it has learned with the ability to comprehend and reason, paving the way for discovery. Now let’s look at the other side of the coin. What has ChatGPT discovered, which nowadays we all have propellers around? Anything.
A program or “machine” is intelligent?
ChatGPT and similar programs are nothing more than highly advanced artificial language processing models that are fed trillions of bits of data. The fact that we get logical, even correct, answers when we ask questions does not mean that a program or “machine” is intelligent. So to say that any language model today has real intelligence can only be a far-fetched characterization.
Being multifunctional: The legendary Garry Kasparov, who was world chess champion from 1985 to 2000 and had an all-time high strength rating of 2851 ELO until 2014, played a match in 1996 with DeepBlue, a chess supercomputer developed by IBM that can analyze 200 million moves per second. He managed to win this match 4-2. His second match in 1997 was controversially lost 3.5–2.5. Kasparov’s success strengthened the claim that humans were still superior to machines in chess at the time.
However, in parallel with the dizzying pace of technology, algorithms running in the background of chess computers have also evolved. Chess computers, which were developed by basically blending reinforcement learning and deep learning, have reached such a level today that they are no longer rivaled.
It’s impossible for any one person to win against them. Even Magnus Carlsen, the strongest chess player of all time, a multiple world champion and the highest ELO ever (2882), cannot rival the Stockfish (3534), which is now considered the most powerful chess computer (publicly available). And that’s not all: let’s remember that Alphazero, developed by DeepMind, had a record of 155 wins, 839 draws and only 6 defeats against Stockfish in a thousand-match series in 2019. Is Chess still a sport or Talent?
What is Real AI? Chess computers are “smart”?
At the same time, let’s not forget that Alphazero can only analyze 60,000 moves per second, against Stockfish, which can analyze 60 million moves per second. Now, the programmers who developed the algorithms of chess computers began to pay attention not only to finding the best move, but also to positional positions, and thus a great breakthrough took place. Moreover, there are no obstacles in front of them for their further development.
Does this mean that chess computers are “smart”? Of course not! Chess computers do nothing but play chess very well. They cannot think and comprehend. They are unconscious. They can only mathematically weight the available possibilities and choose from them. So that’s not a real intelligence. If one day they question why they play chess, then we should start to be afraid.
Appearing Impossible: Even if we mobilize all our possibilities to reveal real artificial intelligence, it is doubtful that we will succeed. If artificial intelligence is not going to be created by human beings, it can only emerge in two ways. The first is in the form of Boltzmann brains.
Powerful anthropic principle
Dedicated to Ludwig Boltzmann, the legendary Austrian physicist and one of the fathers of thermodynamics, Boltzmann brains predict that random quantum fluctuations will eventually create artificial consciousness, or artificial intelligence. Of course, it is thought that a long time must pass from the age of the universe for this, approximately a year.
So long that it was calculated that long before the Boltzmann brains appeared (about a year later), the heat death of our universe would occur. Nevertheless, Boltzmann brains are statistically possible and will surely exist according to the powerful anthropic principle.
It seems impossible for us to intervene in this situation and there seems to be no choice but to wait and see. The second is the result of a more theological approach. A magical “divine” touch. As you can imagine, it is impossible for us to intervene in this situation either. Therefore, the idea that it may be impossible for human beings to create a pure and real intelligence is open to debate.
What is Real AI? So what do we do?
As a result of these three associations, it is now easier to understand that not every automation or algorithm we see can be artificial intelligence.
So what do we do? If we want to create truly conscious, thinking machines, we must urgently change our attitude on this issue and direct artificial intelligence studies to their real goal. We should examine the work of scientists such as Daniel C. Dennett and Douglas R. Hofstadter, who have done serious research on the nature of consciousness, and try to understand and discover “consciousness”, which is perhaps the last mystery we have left.
An example of the most serious work on this subject to date is Douglas R. Hofstadter’s immortal work “Gödel, Escher, Bach: An Eternal Golden Braid”, which was also taught as a course at the Massachusetts Institute of Technology for a while. Similarly, Daniel C. Dennett’s “Consciousness Explained” and “The Mind’s I”, co-authored by the two authors, are excellent books on the subject.
If we can one day clear the smokescreen in front of consciousness, we will have made a great leap forward in reaching true artificial intelligence. If we do not set the goal correctly, the path we take will not matter. Otherwise, we will not be able to go beyond creating more and more advanced Artificial “Missing” Intelligence every day.