Swedish philosopher Nick Bostrom, director of the Oxford Institute for the Future of Humanity, is famous for his work on existential risks, the anthropic principle, the ethics of human improvement and the threats posed by artificial intelligence. He penned "Proof of Simulation", in which he suggests that our reality can be simulated on a computer created by some advanced civilization. In June, Nick Bostrom visited Russia to participate in the International Cybersecurity Congress organized by Sberbank with the support of ANO Tsifrovaya Economy and the Association of Russian Banks. The special correspondent of "Lenta.ru" talked with Bostrom about the uprising of the machines, world government and transhumanism, and also found out why live on if the "Matrix" is around.
What is the meaning of futurology? If we look at the predictions of the futurists of the past, almost none of their predictions turned out to be accurate
Bostrom: Speaking of this discipline, I don't think it has any value.
Nick Bostrom. Photo: Tom Pilston for The Washington Post / Getty Images.
Yes, but you are just trying to predict certain global events in the future
I would not describe as futurology what I or a group of researchers under my leadership do. We do certain things. First of all, we develop concepts to make it easier to find patterns between events in the present and their long-term consequences, to see where they can lead us and whether something can be fundamentally changed.
We also try to find out which changes will lead to a positive outcome, and which to a negative one. For example, the concept of existential risks leading to the fact that intelligent life will die out or get stuck in some kind of radical sub-optimal state. In this example, the consequences of an existential catastrophe are indeed very easy to predict, and they are extremely relevant in terms of a certain value perspective, which implies that the future is of great importance. So all of this will help focus on the consequences of certain actions.
Promotional video:
But some of the research we are doing is much more specific. For example, we have a group of experts working on the security of artificial intelligence (this is a subsection of computer science), because we believe that such insights will definitely be needed in the future.
Should we really be afraid of things like strong artificial intelligence, the fall of a celestial body, the invasion of alien civilizations? Don't you think that when authorities like you, Elon Musk or Stephen Hawking talk about such things, they distract society from more pressing problems?
I do not think that we should only think about the problems that we have now. We should think about where we are going. You need to worry about both the present and the future. Putting on blinkers and not seeing the paths that lead to the cliff would be a big mistake.
Do you really think that artificial intelligence can pose a threat to humanity?
Of course, it is a strong artificial intelligence that may pose a threat - not the one that exists now. I can imagine certain scenarios when this is possible. However, I see them more as a portal through which humanity must pass on the way to a better future, which implies the inevitable development of machine superintelligence. So the question is not whether this step should be taken or not. It's just that when we do it, it should be as careful and responsible as possible. We need to make sure we get through the portal and not crash against the wall.
But does humanity need to go through this portal if a strong AI turns out to be better, smarter, faster than a human? In general, why, in a philosophical sense, is a person needed if a strong AI will represent a more perfect tool for self-knowledge of the universe?
I think we should use machines as an extension of human will, so that they can benefit us. And it is possible. But there are also risks that may prevent us from achieving this goal.
I am trying to ask if there is any sense in the existence of an imperfect organic life if the machine is better? Why would humanity even exist after developing strong AI?
You know, it all depends on the alternatives - what the result will be humanity and what these other forms of life will be. I do not think that if something turns out to be more reasonable, more successful in an evolutionary sense than a person, then from the point of view of morality it will be better than a person. I imagine various outcomes leading to significantly negative consequences: for example, an extremely advanced technological civilization can lose universal values. But I certainly do not believe that humans in their current form are ideal beings with nowhere else to evolve. And here I am not only talking about the improvement of biological life forms - one can imagine positive scenarios in which humanity will become a completely digital form of life.
So you think that consciousness can be transferred to a car?
Yes, but with the advent of mature enough technology, not today.
And, therefore, you believe that consciousness is exclusively the numbering processes?
Yes.
How do you feel about the works of philosophers who postulate the dualistic nature of consciousness - that consciousness is a subjective phenomenon that exists outside the objective world? For example, as the Australian philosopher David Chalmers describes it
I am an adherent of the theory that consciousness is a number. Of course, I do not believe that what makes us conscious beings gives us a subjective experience depends on the fact that we are simply made of carbon atoms. But our brain produces certain numbering processes that, in principle, can be launched in a different environment. And this is the key postulate - it is not the material that is important, but the functional attributes of the numbering process themselves.
And the fact that I perceive the world through my consciousness is just a set of neural impulses in my brain?
Exactly. I believe that your brain gives rise to your consciousness. But the question is what else can generate your consciousness. And I think that in principle it doesn't have to be an organic structure. This could be a computer, for example.
But will he be aware of himself, have that very subjectivity?
I believe it is possible to create self-aware computers. Again, this is not possible now, but in the future I see no obstacles for this.
If he were self-aware, wouldn't humanity have to accept him as an equal intelligent life form?
It seems to me that there may be many completely different digital minds, much more diverse than human ones. Although we believe that we are all very different - Mother Teresa, Hitler, the smartest person in the world and the stupidest - we are all terribly similar to each other, the brain of each of us is not much different from any other human brain. All this is about one and a half kilograms of hundreds of billions of neurons. But digital minds can be very, very different and "alien" to us (some, of course, may be similar to ours, and some - not).
At the same time, I believe that some digital minds (especially those that will have consciousness, will be able to feel pain, pressure, and so on) will be assigned different gradations of moral status. As we develop more and more sophisticated and sophisticated artificial intelligence, the digital mind, we will have to face a certain ethical issue. And not only about how AI will influence us or how one person can apply AI in relation to another person, but also about how we will relate to this digital mind.
Suppose I built a small cairn of stones and destroyed it - I did not harm anyone. If, of course, you build it, and I destroy it, then it may harm you, but the cairn itself has no moral status. And, say, a dog has it, and therefore it is wrong to kick a dog. It's the same with a conscious digital mind, even if at first its consciousness is on the same level as that of an animal. This raises the ethical question of how we should treat it. And I believe that the time will soon come to think about this seriously, not only in the framework of philosophical seminars, but also in a broad public discussion.
Frame: the film "Keepers".
If we talk about global threats, can the increase in the number of authoritarian regimes in the world be considered one of them?
Yes, I think this is one of the possible scenarios of a dystopian future that can be imagined if various sociopolitical trends develop in the wrong direction for long enough. Some advanced technologies may make it easier to develop these trends: better citizen surveillance, more advanced drones …
Now, in order to rule a country, even if you are a dictator, you need the support of the military and elites. You can imagine that with the help of artificial intelligence technology and robotics, the number of people required for this will constantly decrease. In the end, it may turn out that a dozen people will be able to rule the population of a large country, and this provides rich opportunities for maintaining the existence of despotic regimes.
So yes - this is one of the risks, but there is also the possibility that these sociopolitical trends will go in a different direction.
You often talk and write about the singleton - a state association in which decisions are made by a single world body …
If you look into the history of our species, then initially the largest unit of political integration was collectives of hunter-gatherers, consisting of a maximum of fifty people. Over time, tribes were formed under the leadership of leaders, then city-states, then nation-states. Now we have associations like the European Union, various international organizations, and this will continue until the singleton appears. That is, it is a continuation of long-standing historical trends.
As I understand it, you postulated that there could be two types of singleton: “bad” - authoritarian and “good” - egalitarian …
I do not at all consider a "bad" singleton necessarily authoritarian, and a "good" one - egalitarian. There are various circumstances in which a singleton can be "good" or "bad."
Explain, please
I don't even know how to summarize it all … But, of course, you can imagine a "bad" egalitarian singleton. Let's say he may be incompetent or ineffective for various reasons. For example, if everyone lives equally badly, that would be a completely egalitarian outcome, wouldn't it? But still bad.
So there is a possibility of a "good" authoritarian singleton emerging?
Well, perhaps, when compared with the various alternatives, in a situation where a well-meaning dictator is at the helm … “Good” and “bad” are relative terms used in view of other possible situations. Say, if the alternative is a global catastrophe, that outcome would be relatively good.
But what if you imagine the perfect egalitarian singleton? All this looks like a kind of utopia in which people live without conflicts and wars in complete harmony. But conflicts are inherent in human nature
Human nature is not cast in bronze. If we consider the situation in which the technological maturity of civilization was achieved (here I mean both biotechnological and cultural means of changing its various aspects), if the goal of achieving the common good is set, this can be a good reason for modifying human nature.
And how exactly can this be achieved? Become less aggressive, more accommodating, and so on? After all, people tend not to trust strangers, to group themselves into associations that oppose other associations
Yes, and that is their choice and can be changed. But if one group of people treats another badly, then they may decide not to change in such a way as to find common ground with another.
That is, it is most likely that human behavior cannot be changed on a global scale?
Exactly. Here we are talking about a change in motivation, and before that I talked about the possibility of such a change as technological opportunities arise. And then, of course, the question arises whether people will decide to use them to achieve this goal.
By the way, about technology. We are always afraid of new technologies. For example, I recently read a column in an American publication, the author of which campaigned against the abandonment of cash paper money, because non-cash payments allegedly give too many opportunities for banks to control the population. Do you agree with this point of view? Is this really a threat to society?
Cash is an anonymous security. You can pay with banknotes and not be identified. All this reduces the ability of banks and the state to control the daily life of citizens, which under certain circumstances can be quite useful - for example, in despotic regimes. But for the same reasons, it opens the way for all kinds of negative transactions. Criminal activity involves anonymous payments, transfer of bills in envelopes or their modern counterpart - payments in cryptocurrencies. They are much more difficult to carry out, say, through the VISA system, where your personal data is linked to each transaction.
So it is very difficult to decide in which direction to "move the slider". There are different scenarios in which you might want to have more social control over transactions in order to make life harder for criminal organizations, for example. Since their activities are harmful to society, you will most likely want to have ways to restrict them, and one such way is to regulate financial transactions.
This is one example. But what about activists who are calling for the ban of Uber and similar services, since they, they say, violate the rights of taxi drivers?
I don't know what rights Uber is violating. I think they just don't want competition.
Do you think such people can be called neoluddites?
You need to look at the argumentation of specific individuals. In any case, I do not think that labeling them will lead to an understanding of their position. You need to listen to their arguments and decide whether they are good from your point of view or not.
Frame: the film "Equilibrium".
Let's talk about transhumanism. There is a popular criticism of transhumanism, according to which biotechnology, body modifications will not be available to poor people and citizens with average incomes. Only the rich - the very one percent of the population - can get them, and this will lead to a catastrophic increase in inequality in society. This is not true? Or do you think everything is fair: the rich and successful should have the best?
This needs to be analyzed with specific examples. I do not believe that all biotechnology used to improve the human body should be expensive. Take, for example, modern-day Modafinil, a drug used to improve cognitive performance (let's say it helps some people). Sure, it costs money, but no more than a glass of Starbucks cappuccino they drink every day. And if he does his job, he is worth his money.
Modern pharmacological agents that somehow improve certain properties of a person are generally not very expensive. Rather, the problem is that you need to get prescriptions for them. It is easier to get such a drug for a person with a good education. First of all, he will know about the existence of this remedy, it will be easier for him to find a doctor he knows who can write him this prescription … All this will be much more difficult for an uneducated worker or peasant to do than for an intellectual, a banker, and so on. But the obstacle lies in this, and not in the amount of money itself.
Of course, if we talk about the technologies of the future, they can be very expensive and discriminatory in this sense. And society will have to decide if they are useful enough to subsidize them - in the same sense that subsidies for health care are now provided.
If we talk about subsidies, what do you think about the guaranteed income, the amount of money regularly paid by the state to each member of society, regardless of whether he works or not? Should it become a common human practice?
I think that this problem should be approached from the same point of view as artificial intelligence: it should benefit all of humanity and comply with ethical ideals. It seems to me that all people who reach a certain age should receive their share of certain benefits. How best to do this, it will be a constant stream of income or a certain capital paid once, which an individual can then spend at his own discretion on certain services and goods, I do not know yet. Maintaining an uninterrupted flow of income, again, depends on the economic situation. And in the long term, it should be understood that most likely we will have to face an exponential growth in the world's population, which may exceed the growth of the world economy - a somewhat Malthusian argument.
Do you think the current European welfare system - Welfer - has flaws?
It seems quite good to me. Of course, there are ways to improve it. But in any case, it's good when there is a welfare system in rich economies that can afford it.
What do you think of contemporary left-wing politics - political correctness, inclusiveness, and so on? Does she really work for the good of everyone?
On the one hand, the presence of things that cannot be talked about and thought about deprives me as a philosopher of some possibilities. I think that sometimes the bitter truth is good, even if it offends someone, and you have to pay a considerable price for tabooing the discussion of what is part of modern reality.
On the other hand, if this contributes to the construction of a more sensitive society to the problems of the community, whose representatives do their best not to harm each other, thus reducing the level of aggression and violence … All this can turn out to be an extremely positive thing, even if it represents certain risks. See, if you are extremely anxious to be politically correct about small day-to-day issues, the likelihood that you intend to carry out some kind of major and brutal ethnic cleansing or start a war is significantly reduced. In general, if it makes us more sensitive in small things, then it will help in a global sense, and all the small shortcomings of such programs will be absolutely justified.
I read in one of your publications that modern Western culture "actively sabotages our fertility due to the introduction of birth control." A fairly conservative argument. Can you explain it?
This is what I meant: if you don't use birth control, you will have more babies, which means more babies will survive. Now people give birth to fewer children than they could, which means that they are not in evolutionary balance with the modern environment.
Should we give the primary role to biological evolution rather than social?
I believe that, first of all, the main role should be played by universal human values, and not some biological tasks.
Then why not use birth control?
Of course, you can sabotage your reproductive potential, but you must be aware of what you are doing, and this is very beneficial. For example, if you are considering long-term scenarios for the development of human civilization, you need to take into account that people who sabotage reproductive potential will somehow disappear from the face of the earth. Their genes will not be in the planetary gene pool unless this policy is changed.
Now, in the short term, you don't need to think about it, but you should be aware that here you are acting against various evolutionary tendencies. And it's important to recognize that this will have some effect in the long run.
Frame: the film "Lucy".
Is biological evolution still working in human society?
Of course. It works in the human population just like it does in other populations. Of course, the modern environment in which we live is very different from the environment of evolutionary adaptability, but I believe that right now there is a very strong evolutionary selection in the human population. On the ecological time scale, it goes quickly, but, say, in relation to the time scale of technological progress, on the contrary, very slowly. Even when selection is very fast, many generations must pass before a big shift occurs. At the same time, I think that in less than a couple of generations we will receive perfect tools for manipulating the human genome and other possibilities for changing the state of a person. Therefore, I am not particularly concerned about the slow trends in biological evolution, given the prospect of such technologies on the horizon.
Should we choose the path of improving the gene pool through biotechnology - selecting the “bad” genes and keeping the “good” ones? Will this not lead to social Darwinism?
I do not think that genetic medicine or genetic selection presupposes a social Darwinian approach. You can see how this is happening now: when couples decide to undergo in vitro fertilization, the embryos undergo genetic diagnostics. Instead of just checking cells for chromosomal abnormalities, we can diagnose a wide range of problems, including those not related to disease. For example, predisposition to a long healthy life, intelligence, athletic ability, personality traits … You will have more information in order to choose which of 6-8-10 embryos to implant during the IVF procedure. And it seems to me that this very context is most likely in terms of the use of genetic technologies.
You are the author of the "proof of simulation" according to which we may be living in a reality simulated on some kind of computer. How can an ordinary person imagine this? It's hard to digest at first
I don't quite understand what task you are setting. You just need to accept that nothing changes, it's just that the basis of this observable reality in which we exist is not the basic laws of physics, but some kind of computer program embedded in a computer, created by some advanced intelligent beings - perhaps more advanced in technological terms civilization (probably superintelligent). So all that we see is the processes running on this computer, but for the inhabitants of this simulation - us - there is no big difference in what is its base.
It is more interesting to think about the practical reasons for creating such a simulation. They can be different, its authors can run other simulations to solve other problems. There are many options for how the simulation might end - and so on. It really matters when compared to a situation where this reality would be basic.
If we talk about a computer that simulates reality, what about the laws of physics? Such a computer will be gigantic and will consume an unimaginable amount of energy, otherwise how to calculate each elementary particle?
Sure. Therefore, I do not think that this civilization will imitate all the details of reality, down to atoms. I believe that it is more rational to imitate those parts of reality that we observe at the time when we observe them (maybe some more).
Take, for example, the atoms of the table you are sitting at. You have no idea what is happening to them now. You only need to know the dimensions of the table, its color, its weight. And in the simulation it will be enough to calculate only these attributes. If you take an electron microscope and put a piece of a table under it, you will see more details that will be specially calculated for this. But they will be calculated only in such a way that you, an ordinary person, will not notice any anomalies.
Thus, if you are the superintelligence that completely controls reality - down to all aspects of its elements, it is wise to add details as and when you pay attention to them.
It turns out that reality is simulated for all seven billion people living on earth? For animals? For aliens?
It all depends on the type of simulation. If we talk about simulation for all living people, one can imagine that it covers the entire history of mankind and was launched quite recently. You can also imagine a simulation being performed for one particular person or a small group of people. Then, of course, the question arises: how do all the other people we see in it are here, while not being conscious beings? Perhaps they are only simulated in terms of their appearance, like the avatars found in video games. One unresolved question remains: how to imitate the external behavior of people without producing, as a side effect, the consciousness that, as we usually assume, each person possesses? So, probably, creating a believable person in terms of behavior,you will need to simulate the processes taking place in his brain with particular accuracy, which in one way or another should lead to the emergence of a conscious mind. Although, of course, the possibility exists that you could simply fake it. And then we can imagine the existence of a small group of simulations-people in which consciousness has developed completely, while the rest are just extras.
According to your hypothesis, there can be many simulations in a simulation - like nesting dolls. Is the existence of a basic, not imitated universe obligatory in this case? How did it appear and what is it?
My argument assumes that there is some basic level of physical reality in which some creatures develop, become extremely technologically advanced, use these technologies to obtain superintelligence (or superintelligent artificial intelligence) and then they have a lot of resources and time. They spend it, including creating these simulations of intelligent people - perhaps like us.
Perhaps they allow the simulation to continue to function after these people create their simulation, which then becomes a second level simulation. But all this will exist on a computer created by a civilization that exists at the basic level of physical reality, and at the same time, there may be a whole hierarchy of simulations in simulation. But here there will be a specific limit to all this, since in the end the imitation is performed on a computer that exists at the basic level of physical reality. There are limits on the available energy to maintain these capacities, and therefore a limit on the number of nested simulations. Of course, you can take some esoteric approach, try to find loopholes for endless nesting, but I don't think it's worth doing.
What if they simulate worlds in which the laws of physics are completely different from ours?
This is also possible, they can create some kind of fantastic simulations.
Well, if we live in a simulation, what is the point of making plans, predicting global risks … Existential question - what is the point of life in a simulation, when its creators can simply turn off the power of the computer and everything will end?
Generally speaking, “proof of simulation” does not mean that we are living in a simulation, but rather about the limits of what we can believe in. It assumes that one of three statements is true. The first is that we live in a simulation, but there are two alternatives. The second is that all civilizations die out before reaching technological maturity, and the third is that all mature civilizations lose interest in creating such simulations, they simply decide not to do them.
But, within the framework of our argument, if we really live in a simulation, I don't see why the meaning of everyday life depends on the fact that the basis of our reality is imitation created by some advanced civilization, and not the basic laws of physics. In any case, what will happen to you next, what will happen to your relatives, friends and other people depends on your actions. So what you do in the simulation matters, even if outside of it doesn't.
In addition, there is a possibility that what happens inside the simulation affects what is outside it. Perhaps the creators developed it for scientific purposes, in order to study what people do, and this affects what they do, because it is important for them to understand human nature based on the choices we make. So there might be an additional set of consequences beyond the obvious.
Interviewed by Mikhail Karpov