Interview with Nikolai Gorkavym
Nikolai Gorkavy - the author of the trilogy "Astrovityanka", agreed to answer a few questions from Alexey Anshakov, the founder of the webRunes team, which is developing the open source platform IA - Individual Assistant.
Nikolai Gorkavy - astrophysicist, writer, doctor of physical and mathematical sciences. He graduated from the Physics Department of Chelyabinsk State University and graduate school at the Institute of Astronomy of the Russian Academy of Sciences. For 12 years he worked at the Simeiz Observatory.
Winner of the 1989 State Prize of the USSR for predicting Uranus satellites. Together with John Mather theoretically predicted exoplanets near Vega and Epsilon Eridani. He currently lives in the United States and works in the data processing group of the Suomi satellite. The asteroid 4654 Gor'kavyj is named after Nikolai Gorky.
Inca is an artificial intelligence described by the author in the last book of the trilogy “The Return of the Astrovityanka”.
')
Disclaimer: in order to avoid suspicion of advertising, we do not give any information about ourselves or the project, please understand - this is not because we are so secretive. But in the future we will share with the Habr community useful information and development experience.
Alexey: Good afternoon, Nikolay. Thank you for agreeing to be interviewed. When I first shared the idea of ​​IA, one of my acquaintances recommended me to read “Astrovityanka” and, already reading the third part, I was inspired by the spiritually close ideas you described regarding artificial intelligence. How do you think, how many people now live on the idea of ​​creating Inca-like technologies and how long do we have to wait for the first versions of cybersystems?
Nikolay: There are a lot of people who are engaged in artificial intelligence. And they have been doing this for a long time. I have on my shelf two thick volumes of "The Application of Artificial Intelligence in Space", it seems, of the 1978 edition. Written off from the NASA library due to uselessness. There have been very few useful results in the field of AI. For the time being, tasks such as optimal choice of a route by cyber, or automatic docking of ships are being solved. But 15 years ago, robots were exotic for eccentrics, and now this is a whole industry, created at the request of the military, doctors and rescuers. So breakthroughs are possible.
Alexey: As Ray Bradbury complained, mankind has forgotten how to dream about the global, about space and is mired in consumerism and petty desires. But how would you rate the idea of ​​exploring and colonizing space using AI these days?
Nikolay: I described in my book a variant of the colonization of the bodies of the Solar System, when robots with AI, which are able to build a base from local materials and without the direct human intervention, are sent forward. And the man arrives no longer in the wilderness, but in a comfortable city. So, for space colonization, AI will probably be more important than powerful rocket engines.
Alexey: Science fiction writers have always been far ahead of progress, because they are limited only by their imagination, and not the development of science. Do you think it is possible to create an AI at all similar to the Inca described by you and, if so, what is the prediction of the appearance of the first AI in time, at least in the key: this half century, second, next century? And what do you think in this light about the Chinese room ?
Nikolay: I believe that it takes about fifty years for the first versions of real artificial intelligence to appear, similar to Inca, with which you can talk without being irritated by his incomprehension or that he does not perceive your speech, although you left the dentist two hours ago , and your frozen jaw is almost thawed out. Such an AI should not be a bastard, but a system capable of independent search and analysis of important information for you, as well as practically useful tips.
The question of the Chinese room is more a philosophical question than a cybernetic one. The trick is that the mind is perceived by us as a kind of qualitative — and so far incomprehensible — leap over the sorting of binary numbers. There is an element of enthusiastic admiration. Self-appreciation, if you will. When we thoroughly understand how the cyber system functions, it is difficult for us to recognize it as reasonable. No secrets! Therefore, we do not want to attribute the intelligence of the Chinese room of any scale. Who knows when we will look into the details - how the brain functions in a person - won't we deny the rationality of the majority of the human race? Just think, a two-legged crowd of Chinese rooms walks with gramophone in his head and with a simple logic of calling behavioral routines.
Alexey: How not to remember the phrase Bulgakov said through the words of Professor Bormental:
- So he said? It does not mean to be human.
Unfortunately, everything that is not invented is used by mankind primarily for military purposes. But from what spheres, in your opinion, would it be worthwhile to start using AI from a commercial point of view, and in which areas for humanity as a whole?
Nikolay: I think it’s commercial time for simple but useful AI. You do not need a speech recognizer and synthesizer, you do not need to aim at a cyberpoet or a cyber psychiatrist - you need to do at least something useful, capable of independent work. For example, now the search engines for given words almost instantly give out a bunch of information that the person himself has to go through. Now imagine a search engine to which a person sets 10-20 keywords, as well as the maximum file size - 10 pages - and he goes to sleep. In the morning it is waiting for the file, which contains fifty of the most interesting paragraphs from different sources, with their list below, as in a regular article. After reviewing this material and the most interesting links, the student in two accounts will make the necessary essay, and the scientist will understand at what level is the topic that interested him.
Or another example. If you take a large database - for example, earthquakes, then it forms, at a minimum, a space of five dimensions - three coordinates, time, and a push magnitude. Let us set a simple task - to compare seismic activity in two rectangular areas of this space and calculate the correlation coefficient between seismicities as functions of time. It is easy to understand that choosing one rectangular area by coordinates and magnitude gives 8 variables. The second area will give 8 more variables. So, in order to iterate over at least 10 values ​​of each variable, you need to make 10 16 (ten million billion) samples from the base and calculate the correlation coefficient. In the year of 30 million seconds. How many years will it take a regular computer to complete this puzzle? We don’t say that the time function can have a different averaging step, but it would be good to check the correlation even with time shifts. How does a person solve this problem? He makes calculations and analyzes all the time - will there be a sign of correlation, which will require a deepening in this direction? And of course, he can look at the right path — the work day is over, and next week it’s time to leave ... If only someone would make an independent data analyzer — even with ant intelligence — that could crawl around the clock for a given loose pile of numbers in search of pearls of patterns and report to a person once a day: “That's what I found, master!”
For humanity as a whole, the most important is the training Internet systems, which would allow every teenager to have a personal teacher who has infinite patience and is ready to adapt to the level of any schoolchild without mocking him.
Alexey: In today's world, a huge number of lonely people who do not understand, do not perceive their shortcomings. And you described the scene in the third part of “Astrovityanka” in which such people found comfort in communicating with the Ynca. Do you not think that over time, many people will seek refuge in communicating with a clever machine, but not trying to build interpersonal relationships that imply such notions as: compromise and tolerance for the shortcomings of a loved one, loyalty, need to look after yourself, think before you say and more? After all, to please others, people often try to be better. And with android all this is not necessary, it will be with you, no matter what you do, you just have to have the necessary amount on the account to purchase it.
Nikolay: The problem of escapism has repeatedly arisen before humanity. People leave from communication with other people in a drunken reading of books. People are recklessly cut into cards, spitting on the family and children. TV sucked most of the free time most of the population. Now the Internet and social networks have given even more opportunities to achieve not real, but virtual success. Yes, a new round of this problem will inevitably arise when AI and even humanoid androids appear. Remember how the heroine "Terminator" looks at the robot playing with her son, and thinks - what a good father! Do not get drunk and do not hurt. Let's hope that the real girls will always remain for most young men more attractive than any dolls.
Aleksey: And, as a continuation of the previous question, will we face the problem of war with cars? After all, the intellect implies self-awareness, what thoughts will a creature have, which over time will become much smarter than any of the living people? Contempt? A pity? Some much more complex feelings that are not even known to us and are unlikely to be understood due to natural human limitations? Is it possible to lay the notorious rules formulated by Isaac Asimov to protect humanity from its own generation?
Nikolay: The statement that “intelligence implies self-consciousness” is still hanging in the air: we do not know what intelligence is because we have not built its working model, and we don’t know what self-awareness is and at what stage of cyber system development it arises. Not easy is the question of such emotions as pity and contempt. How anthropomorphic and related to the work of our hormonal system? How possible is the introduction of human-like emotions into the cybersystem? Will it not be a simple imitation, a Chinese emotional room?
The laws of Asimov in robots will certainly be laid down - a war of cars and now it is easy to arrange, because robots with machine guns are already mass-produced. But among developers there are not enough idiots to put the possibility of spinning out of control into the software of such machines.
Note:
Here is what the Blue Brain Project researchers are saying about consciousness:
“If consciousness arises from a critical mass of interactions, then it may be possible. But we really do not understand what consciousness is, therefore it is difficult to talk about it. ”
Original text:
“If it’s possible to argue for some critical mass of interactions, then it may be possible. It is difficult to say. "
Alexey: Will the machine create a world where everything that can be automated and works as efficiently as possible? What professions will remain? What to do with the huge masses of people who have become unnecessary? Will we face new Luddites?
Nikolay: We have long been living in a world where most of the professions and people are not involved in the production of food, clothing or machine tools. Doctors, lawyers, journalists, auto mechanics, waiters, salesmen, bankers, artists, writers, all the inhabitants of the world of television, cinema, sports, fashion and music with songs - have long been self-serving society, that is, working for themselves. This is due to the fact that one farmer feeds hundreds of non-farmers, and one worker makes pants, cars and refrigerators to the top ten non-workers. If you raise automation to one hundred percent, then just the last workers will join the already numerous rows of talkers and dancers.
Alexey: Nikolay, you touched upon this question in your trilogy, but how do you think, can the machine ever create? Create nontrivial pictures, invent beautiful music, songs, jokes, invent?
Nikolay: Maybe, but why should a person take away these living niches? You can invent without much creativity, just transferring ideas from one area to another and computers will do it, because it is useful. You can make a machine that specializes in abstract painting - and it will surpass many current authors. But why hurt them so much? People after all.
Alexey: Nikolay, and the last question. Perhaps you want to say something or recommend potential developers of "smart systems" who are now reading the interview?
Nikolay: Above, I have already described a couple of intellectual programs that would be good to create. No need to chase the showiness, do something effective, minimally useful to a large number of people. And then people will reach for your intellect!
Alexey: Thank you, Nikolay, for the answers!
And to our readers: if you want Nikolai Gorky to answer your questions, please ask them in the comments. The list of questions will be given to Nicholas, and his answers are published in the next article.
Thank.
UPDATE: Answers to interview questions can be found here.
Nikolai Gorkavy - astrophysicist, writer, doctor of physical and mathematical sciences. He graduated from the Physics Department of Chelyabinsk State University and graduate school at the Institute of Astronomy of the Russian Academy of Sciences. For 12 years he worked at the Simeiz Observatory.
Winner of the 1989 State Prize of the USSR for predicting Uranus satellites. Together with John Mather theoretically predicted exoplanets near Vega and Epsilon Eridani. He currently lives in the United States and works in the data processing group of the Suomi satellite. The asteroid 4654 Gor'kavyj is named after Nikolai Gorky.
Inca is an artificial intelligence described by the author in the last book of the trilogy “The Return of the Astrovityanka”.
')
Disclaimer: in order to avoid suspicion of advertising, we do not give any information about ourselves or the project, please understand - this is not because we are so secretive. But in the future we will share with the Habr community useful information and development experience.
Interview
Alexey: Good afternoon, Nikolay. Thank you for agreeing to be interviewed. When I first shared the idea of ​​IA, one of my acquaintances recommended me to read “Astrovityanka” and, already reading the third part, I was inspired by the spiritually close ideas you described regarding artificial intelligence. How do you think, how many people now live on the idea of ​​creating Inca-like technologies and how long do we have to wait for the first versions of cybersystems?
Nikolay: There are a lot of people who are engaged in artificial intelligence. And they have been doing this for a long time. I have on my shelf two thick volumes of "The Application of Artificial Intelligence in Space", it seems, of the 1978 edition. Written off from the NASA library due to uselessness. There have been very few useful results in the field of AI. For the time being, tasks such as optimal choice of a route by cyber, or automatic docking of ships are being solved. But 15 years ago, robots were exotic for eccentrics, and now this is a whole industry, created at the request of the military, doctors and rescuers. So breakthroughs are possible.
Alexey: As Ray Bradbury complained, mankind has forgotten how to dream about the global, about space and is mired in consumerism and petty desires. But how would you rate the idea of ​​exploring and colonizing space using AI these days?
Nikolay: I described in my book a variant of the colonization of the bodies of the Solar System, when robots with AI, which are able to build a base from local materials and without the direct human intervention, are sent forward. And the man arrives no longer in the wilderness, but in a comfortable city. So, for space colonization, AI will probably be more important than powerful rocket engines.
Alexey: Science fiction writers have always been far ahead of progress, because they are limited only by their imagination, and not the development of science. Do you think it is possible to create an AI at all similar to the Inca described by you and, if so, what is the prediction of the appearance of the first AI in time, at least in the key: this half century, second, next century? And what do you think in this light about the Chinese room ?
Nikolay: I believe that it takes about fifty years for the first versions of real artificial intelligence to appear, similar to Inca, with which you can talk without being irritated by his incomprehension or that he does not perceive your speech, although you left the dentist two hours ago , and your frozen jaw is almost thawed out. Such an AI should not be a bastard, but a system capable of independent search and analysis of important information for you, as well as practically useful tips.
The question of the Chinese room is more a philosophical question than a cybernetic one. The trick is that the mind is perceived by us as a kind of qualitative — and so far incomprehensible — leap over the sorting of binary numbers. There is an element of enthusiastic admiration. Self-appreciation, if you will. When we thoroughly understand how the cyber system functions, it is difficult for us to recognize it as reasonable. No secrets! Therefore, we do not want to attribute the intelligence of the Chinese room of any scale. Who knows when we will look into the details - how the brain functions in a person - won't we deny the rationality of the majority of the human race? Just think, a two-legged crowd of Chinese rooms walks with gramophone in his head and with a simple logic of calling behavioral routines.
Alexey: How not to remember the phrase Bulgakov said through the words of Professor Bormental:
- So he said? It does not mean to be human.
Unfortunately, everything that is not invented is used by mankind primarily for military purposes. But from what spheres, in your opinion, would it be worthwhile to start using AI from a commercial point of view, and in which areas for humanity as a whole?
Nikolay: I think it’s commercial time for simple but useful AI. You do not need a speech recognizer and synthesizer, you do not need to aim at a cyberpoet or a cyber psychiatrist - you need to do at least something useful, capable of independent work. For example, now the search engines for given words almost instantly give out a bunch of information that the person himself has to go through. Now imagine a search engine to which a person sets 10-20 keywords, as well as the maximum file size - 10 pages - and he goes to sleep. In the morning it is waiting for the file, which contains fifty of the most interesting paragraphs from different sources, with their list below, as in a regular article. After reviewing this material and the most interesting links, the student in two accounts will make the necessary essay, and the scientist will understand at what level is the topic that interested him.
Or another example. If you take a large database - for example, earthquakes, then it forms, at a minimum, a space of five dimensions - three coordinates, time, and a push magnitude. Let us set a simple task - to compare seismic activity in two rectangular areas of this space and calculate the correlation coefficient between seismicities as functions of time. It is easy to understand that choosing one rectangular area by coordinates and magnitude gives 8 variables. The second area will give 8 more variables. So, in order to iterate over at least 10 values ​​of each variable, you need to make 10 16 (ten million billion) samples from the base and calculate the correlation coefficient. In the year of 30 million seconds. How many years will it take a regular computer to complete this puzzle? We don’t say that the time function can have a different averaging step, but it would be good to check the correlation even with time shifts. How does a person solve this problem? He makes calculations and analyzes all the time - will there be a sign of correlation, which will require a deepening in this direction? And of course, he can look at the right path — the work day is over, and next week it’s time to leave ... If only someone would make an independent data analyzer — even with ant intelligence — that could crawl around the clock for a given loose pile of numbers in search of pearls of patterns and report to a person once a day: “That's what I found, master!”
For humanity as a whole, the most important is the training Internet systems, which would allow every teenager to have a personal teacher who has infinite patience and is ready to adapt to the level of any schoolchild without mocking him.
Alexey: In today's world, a huge number of lonely people who do not understand, do not perceive their shortcomings. And you described the scene in the third part of “Astrovityanka” in which such people found comfort in communicating with the Ynca. Do you not think that over time, many people will seek refuge in communicating with a clever machine, but not trying to build interpersonal relationships that imply such notions as: compromise and tolerance for the shortcomings of a loved one, loyalty, need to look after yourself, think before you say and more? After all, to please others, people often try to be better. And with android all this is not necessary, it will be with you, no matter what you do, you just have to have the necessary amount on the account to purchase it.
Nikolay: The problem of escapism has repeatedly arisen before humanity. People leave from communication with other people in a drunken reading of books. People are recklessly cut into cards, spitting on the family and children. TV sucked most of the free time most of the population. Now the Internet and social networks have given even more opportunities to achieve not real, but virtual success. Yes, a new round of this problem will inevitably arise when AI and even humanoid androids appear. Remember how the heroine "Terminator" looks at the robot playing with her son, and thinks - what a good father! Do not get drunk and do not hurt. Let's hope that the real girls will always remain for most young men more attractive than any dolls.
Aleksey: And, as a continuation of the previous question, will we face the problem of war with cars? After all, the intellect implies self-awareness, what thoughts will a creature have, which over time will become much smarter than any of the living people? Contempt? A pity? Some much more complex feelings that are not even known to us and are unlikely to be understood due to natural human limitations? Is it possible to lay the notorious rules formulated by Isaac Asimov to protect humanity from its own generation?
Nikolay: The statement that “intelligence implies self-consciousness” is still hanging in the air: we do not know what intelligence is because we have not built its working model, and we don’t know what self-awareness is and at what stage of cyber system development it arises. Not easy is the question of such emotions as pity and contempt. How anthropomorphic and related to the work of our hormonal system? How possible is the introduction of human-like emotions into the cybersystem? Will it not be a simple imitation, a Chinese emotional room?
The laws of Asimov in robots will certainly be laid down - a war of cars and now it is easy to arrange, because robots with machine guns are already mass-produced. But among developers there are not enough idiots to put the possibility of spinning out of control into the software of such machines.
Note:
Here is what the Blue Brain Project researchers are saying about consciousness:
“If consciousness arises from a critical mass of interactions, then it may be possible. But we really do not understand what consciousness is, therefore it is difficult to talk about it. ”
Original text:
“If it’s possible to argue for some critical mass of interactions, then it may be possible. It is difficult to say. "
Alexey: Will the machine create a world where everything that can be automated and works as efficiently as possible? What professions will remain? What to do with the huge masses of people who have become unnecessary? Will we face new Luddites?
Nikolay: We have long been living in a world where most of the professions and people are not involved in the production of food, clothing or machine tools. Doctors, lawyers, journalists, auto mechanics, waiters, salesmen, bankers, artists, writers, all the inhabitants of the world of television, cinema, sports, fashion and music with songs - have long been self-serving society, that is, working for themselves. This is due to the fact that one farmer feeds hundreds of non-farmers, and one worker makes pants, cars and refrigerators to the top ten non-workers. If you raise automation to one hundred percent, then just the last workers will join the already numerous rows of talkers and dancers.
Alexey: Nikolay, you touched upon this question in your trilogy, but how do you think, can the machine ever create? Create nontrivial pictures, invent beautiful music, songs, jokes, invent?
Nikolay: Maybe, but why should a person take away these living niches? You can invent without much creativity, just transferring ideas from one area to another and computers will do it, because it is useful. You can make a machine that specializes in abstract painting - and it will surpass many current authors. But why hurt them so much? People after all.
Alexey: Nikolay, and the last question. Perhaps you want to say something or recommend potential developers of "smart systems" who are now reading the interview?
Nikolay: Above, I have already described a couple of intellectual programs that would be good to create. No need to chase the showiness, do something effective, minimally useful to a large number of people. And then people will reach for your intellect!
Alexey: Thank you, Nikolay, for the answers!
And to our readers: if you want Nikolai Gorky to answer your questions, please ask them in the comments. The list of questions will be given to Nicholas, and his answers are published in the next article.
Thank.
UPDATE: Answers to interview questions can be found here.
Source: https://habr.com/ru/post/151075/
All Articles