The Limited Times

Now you can see non-English news...

Brutal technology that will take over the human race? We will become artificial intelligence Israel today

2022-09-24T10:40:21.648Z


A deep fake that will endanger democracies? "Humans have always cheated" • The day when computers will achieve higher abilities than humans? "It's a simplistic idea, intelligence is a multidimensional concept" • And are we approaching the day when codes will be self-aware? "The road to get there is still long" • From his seat in Oxford, as head of the chair for public understanding of science, Prof. Marcus de Sotoy refutes some conventions about bots, robots and everything in between


One morning, on his way to work, Prof. Marcus du Soutoy listened to a masterful piece of classical music played on the radio.

Du Sotoi, an English professor of mathematics, head of the Department of Public Understanding of Science at the University of Oxford and a lover of classical works, bet his heart on the German composer Johann Sebastian Bach.

The melody of the piece ended, Du Sotoy's car continued to drive down the road, and at the end of the piece a surprise awaited him that surprised him: the narrator discovered that it was a piece written by an algorithm.

But the really surprising fact for du Sotto was not that he was deceived into thinking that it was Bach, but that he was moved by what he heard.

"I was curious," he says, "how the algorithm responsible for the piece tricked me into thinking that the great Bach composed it."

And this is perhaps the central essence of Du Sottoy's new book, "The Creativity Software - How Artificial Intelligence Learns to Write, Draw and Think" (Attic Books and Yediot Books).

It attempts to answer some of the most burning questions in the expanding and evolving world of artificial intelligence (AI): Will a computer ever be able to compose a symphony, an award-winning novel, or paint a masterpiece?

And can machines be creative and deeply understand human complexity to create unique art and even help us expand our human creativity?

"Either we will understand the code, or the code will understand us."

Du Sotoy, photo: Oxford University Images - Joby Session

The answers to these questions are sometimes more complex than it seems, but if we go back to that innocent journey where Du Sottoy listened to the "Unknown Composer", then in summary the piece he listened to that day was created by obeying fairly simple programming rules, which the composer David Cope devised when he was looking for inspiration and turned to to algorithms in a step of desperation.

This algorithm worked well for du Sotoi, but despite the feelings that arose in him - he is in no hurry to get out of his way.

"After all," he says, "algorithms compose in rooms sealed off from emotion, without coming into contact with the world around them. Without a load of experiences their music will not compare to that of the great composers."

So the next historical artist on the scale of Bach or Mozart, and also Shakespeare or Rembrandt - will not be an outgrowth of artificial intelligence?

"She will be there to assist. We are already seeing a new and powerful artist emerging, but it is a combination, a fusion, of human and artificial intelligence. I think this will help us move away from the idea of ​​art being created by an isolated genius. I like the concept of the musician and theorist Brian Eno's music. He coined the term 'community genius' (Scenius) as opposed to individual genius (Genius) to recognize the influence of the community within which creative intelligence grows, as opposed to the concept of the isolated genius as we saw in the past examples of Mozart or Shakespeare."

Running evolution forward?

Du Sotoi (57), a declared atheist and popular science writer, is involved in the popularization of mathematics and researches, along with artificial intelligence and creativity, set theory and number theory.

In 2001 he won the Brick Prize for the best research for a mathematician under 40, and among his previous books you can find "The Music of the Prime Numbers".

In the present book it is evident that Du Sotoy tends to cool down the possibility of artificial intelligence progressing in such a way that one day it will completely replace humans.

While he describes how artificial intelligence may be useful even on the football field, he is not at all sure that we are getting close to the day when it will replace people in writing, public speaking and even psychotherapy.

"We are creating an amazing new tool, artificial intelligence, which will allow us to see into the depths of the data of the digital world and discover new things. We are like Galileo and the telescope, only we explore the digital world and not space. I don't think we will see writing, public speaking or psychotherapy, and not even Football players are controlled by artificial intelligence, but it may help us improve our abilities in these areas."

Do you attribute these limitations of artificial intelligence to the lack of knowledge of their finality and the absence of consciousness?

In the book you make an interesting point, that machines don't know they are going to die.

How does this understanding affect her development and her creativity?

"For me, creativity is humanity's most powerful force to understand our consciousness. Consciousness comes with a heavy price because it allows us to take a mind trip through time, thus imagining ourselves in the future. But this trip involves the understanding that one day we will die. My iPhone does not know that it will become obsolete In a period of two years. But when he becomes aware of this, will he feel the need to leave behind something that will provide proof of his very existence?

"Since artificial intelligence does not yet have consciousness, creativity is not driven by artificial intelligence. When artificial intelligence creates - we still associate the intention with a person. Artificial intelligence did not want to play 'go', it was a person who made it play - Lee Sadol. But what is interesting is That the code that emerges in this new artificial intelligence, even though it is not conscious, is so complex that we have difficulty understanding how it makes its decisions.

"Asking an artificial intelligence to express itself through creativity - visual, musical or verbal - can help us get a glimpse into its complex inner world where we can understand how it thinks and makes decisions. But when an artificial intelligence becomes conscious, it will have an intention because it will be desperate to tell us that there is It has an inner world. This is also the moment when I believe we will suddenly see a huge wave in the creativity of computers. But as long as machines have no consciousness, I don't think they can be more than a device for expanding human creativity."

Are we approaching such a day when artificial intelligence will have awareness?

And if so, how different will it be from human awareness?

"Despite that Google employee's claim that his code has become self-aware, I think we're still a long way from conscious code. But that doesn't mean we should laugh at that Google employee for his belief. He raises an interesting challenge here, for example, in relation to the question of how to tell about anything at all Such. The work I've done on consciousness and expanded on in my previous book, What We Can't Know, suggests that the key or solution may lie with the integrated nature of the brain network. I see no reason why silicon won't achieve it. But the question is how long it will take Can we fast forward evolution?

"Furthermore, in order to build a machine with consciousness, we will need a combination of knowledge from different fields of science. The moment we succeed in this task, I predict that the machine's consciousness will be very different from ours, and at first we will not even realize that it has consciousness. I also believe that it will want, as mentioned, to share her experiences with us."

Instill empathy into the machine

As for the "artificial consciousness", if you will, Du Sottoy claims that the best way to understand how, for example, his iPhone feels is to use storytelling and not technological scanners.

According to him, the ability to share our worlds of consciousness through stories is the feature that, among other things, makes us human.

According to him, sharing stories may save us from implementing horror scenarios related to artificial intelligence.

In the book, he cites as an example the words of the British writer Ian McEwan after the September 11 attacks: "If the hijackers had been able to imagine the thoughts and feelings of the passengers, they would not have been able to continue. It is difficult to behave cruelly when the person allows himself to enter the victim's head. The ability to imagine how the other feels Not being you is at the core of our humanity. This is the essence of compassion and this is the beginning of morality."

Du Sottoy believes that "if machines are made to have consciousness, introducing empathy into the machine could save us from the story of the Terminator from 'Terminator'".

From reading your book and your words, I understand that you are not a follower of the singularity theory, which hypothesizes that at a future point in time computers will achieve intellectual abilities significantly higher than those of humans.

"I think the idea of ​​the singularity is overly simplistic. It suggests that intelligence is a one-dimensional concept. When will artificial intelligence be more intelligent than us? What will be the tipping point? Intelligence is a complex and multidimensional concept. Artificial intelligence is already more intelligent than us in ways Many, while in other ways - humans will always be superior to her.

"Hollywood likes to present the narrative of a brutal artificial intelligence that will take over the human race. The message of my book is that we need to think about cooperation, not competition. There is a lot of evidence that artificial intelligence and humans together are better than either alone. Our future will be is built on a hybrid existence. We will become artificial intelligence."

AI or cheap marketing?

Du Sotto's integrative prediction about our becoming artificial intelligence can be taken in many directions, but according to him, given our co-existence with artificial intelligence, the key to survival in the new and emerging world is a basic understanding of what artificial intelligence and algorithms actually are.

"Artificial intelligence and algorithms are just clever parts of a wider mathematics," he explains.

"Therefore, in the book I try to lift the hood of this engine and show the readers how it works. Only then can you simplify and make the magic accessible. That is why it is very important that our society is digitally literate.

"There is a tremendous enthusiasm for artificial intelligence that we need to be careful about. Almost everything today comes with a piece of artificial intelligence, but mostly it is marketing, and most things are still done by people. That is, artificial intelligence pushes and pulls us, helps us make good decisions and choices More, but we have to be careful. Either we program and understand the code, or we are programmed and the code understands us."

In the book you also wonder if humans will continue to do mathematics.

What is your concern about it?

"In the book, I examine my creative process regarding mathematics and wonder if artificial intelligence will make me unemployable in the future. After all, artificial intelligence is just smart math algorithms. So can't it easily do what I do? But that's one of the things I'm trying Demonstrate in the book: Mathematics is a human, creative and intuitive process that is difficult to automate.

"Recently there was a very interesting development where artificial intelligence found new potential mathematical patterns and proposed new hypotheses. But it was not able to prove them. It already depended on a human being. But again, what we are seeing is a powerful collaboration. Together, artificial intelligence and humans They made progress that would not have happened without joint forces."

In view of the powerful collaboration that Du Sottoy talks about, I ask him about the most advanced artificial intelligence today, and what uses are made of it.

"The most advanced artificial intelligence to come out of the machine learning revolution is in the visual field. The ability of artificial intelligence to recognize and create visual products has shaken the industry. The place where it has the most difficulty is the language field.

"GPT3 is an amazing text generation algorithm. Google Translate allows people to have romantic relationships with people who don't speak their language. But still, I think it's sometimes easy to tell that you're talking to an AI.

"Artificial intelligence systems that try to deal with natural language have difficulty dealing with ambiguity and context. The Lubner test often ends with a test of machine intelligence called 'Winograd challenges,' named after the Stanford professor who devised them. These challenges quickly identify failing speech programs In understanding structured ambiguity.

"The interesting point for me is that language is not just about words, but related to culture, politics, history and music. Artificial intelligence is sometimes given one type of information. GPT3 is trained on the written word, but not on all the information we tap from to understand language. Our intelligence is broad, general and embodies many layers within it".

Writing that serves democracy.

book cover,

To the moon, and beyond

Although our intelligence is broader than that of artificial intelligence in many areas, it seems that more than one technology succeeds in undermining us.

This, for example, was the case that happened recently with the Israeli researcher Dr. Roi Cezana, who "faked" the landing on the moon using artificial intelligence, but at the end of a post he uploaded to the Internet he discovered that it was a fake, created using artificial intelligence engines available on the Internet.

In accordance with what Dr. Cezana tried to show, I ask Du Sotoi if we are entering a period when it will no longer be possible to trust the sight of the eyes and the sound of the ears; The existing order in our world and even to challenge democracies.

"The world has been shaken by the ability of artificial intelligence to produce a deep fake. This is a significant strength of machine learning-based artificial intelligence. Give it a huge amount of information - and it will learn the basic patterns and the unique style hidden in the information, so that it can itself produce works in the same style, whether it is a fake work of Rembrandt And between it being a politician who 'says' things he never said.

"But you have to remember that humans have always been very easily deceived by the sight of the eyes and the sound of the ears. This is one of the reasons why magic and illusions work. Just as artificial intelligence is good at creating fakes, it is also much better than us at detecting fakes, lies. Take for example our emotions - when a smile Is it real and when is it fake?

"It turns out that artificial intelligence is much better at picking up the signals that indicate a fake smile on a human face compared to the ability of humans to recognize it. This may be very effective for children on the spectrum, who show difficulty in reading the other's emotions. Artificial intelligence is able to help with this. This is perhaps the point - intelligence Artificiality is very much a double-edged sword. It can be used for good or bad. It's up to us, as a society, how we use it.

"This is partly why I wrote the current book: to empower people to make decisions about the future of artificial intelligence in our society. This is important for democracy, otherwise we may be led by algorithms. We saw an example of this in the case of Cambridge Analytica - how algorithms are used to influence election results".

Towards the end, and long before using algorithms for the purpose of influencing elections, Du Sotoy illustrates to me in a small way how he himself makes use of the capabilities of artificial intelligence.

He admits that he did not compose the entire book himself, and that he was unable to refuse an offer on behalf of a contemporary version of "The Automatic Grammarian" by Roald Dahl.

It is a 350-word segment written by an algorithm that specializes in producing short articles, based on a few keywords that are entered into it.

I'm curious to hear from Du Sotoi what that part is.

"I was absolutely amazed that only one person was able to correctly identify the AI ​​paragraph in my book. I thought it was pretty obvious because it was written quite badly. What does that say about my writing?"

He is joking.

"I was very strict with my editor that the text would not be corrected in any way, so that the paragraph would remain exactly as it was written. I wonder if the Hebrew version made the task more difficult because the temptation is to make the text as natural as possible. In any case, I do not intend to find out which paragraph it is And ruin the fun of discovery."

On the way to the heights or the edge of existence?

Shaul ben Shushan

To the statement of Yogi Bera, coach and American baseball player, "The future is not what it used to be", it can already be added that neither is the present.

The growing dependence on artificial intelligence (AI), the dramatic improvements in its performance and the predictions that it will replace humans in many fields are flooding the question of whether there is a field where the human advantage will be maintained.

According to Marcus de Soutoy, professor of mathematics at the University of Oxford, this area is creativity, which "millions of years of evolution have polished inside our brains."

As befits a mathematician, he moves from the hypothesis phase to the proof phase and takes us, heel to toe, on a mental journey aimed at testing whether the AI ​​compares to the "human software" and perhaps surpasses it.

When he is careful not to be too didactic, he spices up the explanations with personal stories such as his personal anxiety that the algorithms will rob him of the practice of mathematics, with quotes ("The main enemy of creativity is common sense") and instructive anecdotes - what does "Netflix" do to hit the taste of the viewer?

About a book whose price was raised by an incorrect algorithm to about 1.5 million dollars, about political activists who made it so that every time you googled the word "idiot" Trump's picture appeared, and more.

Du Sotoy attributes the turning point in AI to the "Alphago" software, which at its peak (in 2016) defeated the then world champion in the Chinese game "Go", which is considered more complex than chess.

The change she created is that instead of teaching the software to play, the software writes software for the game, "that will learn from its own mistakes, just as a baby is not programmed to deal with life but to learn and deal with it while interacting with it".

Can mathematical proofs performed by an algorithm be trusted?

- This is the question to which the mathematician Du Sotoi pays special attention.

He extols the advantages of the computer in this field and estimates that "we are reaching the limit of the capacity of our piece of human hardware".

Music - "the expression of mathematics in sounds", as he put it, has a special potential for algorithmic application.

He analyzes, from a mathematical point of view, the structure of Bach's music, wonders whether it is possible to "recreate" his works, and later brings testimonies of artists who the computer helped them break through the barriers of creation.

Throughout the book, he works on both fronts at the same time: on the one hand - describing the operation of the algorithms and the ways in which man breaks his limits with their help, and on the other hand - the threat and growing dependence on them, forcing us to shelve the ultimate solution of cutting off the electricity on the Day of Judgment, since it is no longer an alternative.

Du Sottoy is determined to encompass all the "logarithmic competitors" of human creativity and is guilty of "multiple grasping".

The feeling of déjà vu is inevitable when it comes to writing poetry, fiction and language translations.

Although this approach suits his stated goal of testing and "bringing artificial intelligence to the limit", it seems that this limit goes a little beyond the reader's ability.

The reason for this is that at a certain point the reader realizes that a review of other fields, which are interesting in themselves, cannot advance his basic understanding of whether the creativity of machines is similar or approaches human creativity.

It seems that the lack of a deeper philosophical discussion - at the beginning of the book - about free will, awareness and the self, and their connection to creativity, could have given the reader the necessary breathing space and provided him with a broader and more interesting frame of mind.

This kind of discussion is brought to the fore in the last chapter, "Why do we create?", too little too late.

The fundamental questions, it must be emphasized, do not disappear from the author's eyes, but he addresses them sparingly - with short personal reflections and quotations - perhaps to make the subject accessible and preserve the reader's energy.

Despite this, Du Sottoy succeeds in bringing us the world that a moment ago we were afraid to face.

Attacking the subject from so many angles, his impressive knowledge and ability to tell a story succeed in knocking down the walls that seemed impassable until now.

In the first chapters, he gives the feeling that "the issue is under control", but as you get into the thick of the beam ("man approaches the point where he will be obliged to use the machine"), the reader is gripped by a panic that causes him to cooperate and show activity in his search for an answer.

The solidarity that is created with the writer becomes, to a large extent, the brotherhood of human beings to find their salvation, the thing that cannot be studied and expressed through a series of numbers or patterns, and if you want - you are allowed to be human.

Theorist Neil Postman argued that man has removed responsibility from himself and made technology a goal instead of a means, thus also making all the old values ​​and traditions irrelevant.

Will this technology elevate him to the heights or push him to the brink of existence?

In any case, the singularity - the point where the intelligence of the machines will be immeasurably higher than ours - may occur by 2045, according to Ray Kurzweil - inventor, thinker and futurist.

There is still some time.

have fun

were we wrong

We will fix it!

If you found an error in the article, we would appreciate it if you shared it with us

Source: israelhayom

All news articles on 2022-09-24

You may like

News/Politics 2024-03-23T00:35:34.817Z
News/Politics 2024-03-22T23:04:48.727Z

Trends 24h

News/Politics 2024-03-28T06:04:53.137Z

Latest

© Communities 2019 - Privacy

The information on this site is from external sources that are not under our control.
The inclusion of any links does not necessarily imply a recommendation or endorse the views expressed within them.