Short summary - 2062: The World That AI Made - Toby Walsh

British literature summaries -

Short summary - 2062: The World That AI Made
Toby Walsh

VERY BRIEFLY

The artificial intelligence specialist represents a future in which machines become intellectually equal to people, and tries to answer the question whether they will displace a person.


Homo digitalis

Man built the Great Wall of China and the Egyptian pyramids, conquered the fire and went into space ... But, despite all the achievements, we will be replaced. Almost all traces of Homo sapiens will disappear, just as the traces of Neanderthals who could not resist us disappeared. We will be replaced by artificial intelligence (AI), which will reach our level by 2062.

Homo sapiens supplanted Neanderthals due to their ability to speak. Before speaking, our ability to learn was very limited, but language changed everything. The ancestors learned to transmit to each other the information necessary for survival. The language gave us stories, myths, religions and added sapiens (sensible) to the name of our species. Our strength has become collective.


The second spurt occurred when writing appeared. It originated in China about 5,000 years BC. e. and in Mesopotamia about 3 100 years BC. e. It made it possible to complicate the structure of societies, systematize the laws governing life, and introduce educational systems.

The third step is printing. Gutenberg invented the printing press around the year 1440. The following year, a little less than 100,000 books were printed in Europe. Today, the printing industry publishes millions of books every year.

The fourth step that humanity is taking right now is what the author calls coloring. Coloring is an individual training within a group, while the knowledge of each member of the group is the same. The computer code has become an ideal tool for this type of training, and computers are trained using this method. Computers excel in learning ability. People are limited in means and methods of transferring knowledge, while computers exchange codes with each other without restrictions. If one computer learns to diagnose skin cancer, all other machines will pick up this ability. This is colorizing.

While computers do not surpass us in everything. Our brain is still more complex than the most powerful supercomputer. We are creative. But, perhaps, we will not have these advantages for long.

Our end

Machines beat man in chess and go, but they surpassed us not only in games. They better analyze the Internet to predict stock prices, and in medicine, computers read cardiograms better than doctors. However, this is the ability to solve one specific problem.

The goal of developing artificial intelligence is to write a program that can do everything just like a person, or better. This is still a long way off - people are ahead of cars so far: they are fast-trained and know how to explain their decisions. Algorithms can recognize a cat in the photo, but they cannot say how they determined it. AI does not have a complete picture of the world: it does not associate an apple with the law of gravity. People know how to adapt - this has made us the dominant view of the planet.

According to the author, the "invasion of machines" is highly advertised. So far, AND is not the worst threat to humanity: a survey of 50 Nobel laureates has shown that the AI problem is losing to climate change, population growth, nuclear war and Trump.

End of consciousness

We have no tools to measure consciousness. As far as we know, not a single part of the brain is responsible for it. Animals have a limited level of consciousness, but not AlphaGo. She won’t think: «It has worked out well since I earn some money on online poker.» AlphaGo does not even know that he plays go.

But there is no guarantee that the situation will not change. The author offers three scenarios in which machines can gain consciousness:

  1. It can be programmed.
  2. It will arise as a result of computer improvements.
  3. It can be learned by them.

The first way is the least likely: how to program something that we ourselves do not understand? Perhaps in 2062, AI would have no consciousness at all. Nature reminds us that AI can have a completely different mind. Look at the octopuses - the most intelligent of all invertebrates.

Even if the AI goes unconscious, there will remain an equally complex issue of free will.

Finally, when we try to make a digital copy of our biological mind, we run into technological limitations: the brain with its billions of neurons and trillions of synapses is the most complex system in the universe.

End of work

In 2062, people will work much less. Robots will work, and we will do more pleasant and important things - we will create art, develop society ... This will be the Second Renaissance! Or not?

Economists worry that by 2062, most professions will be killed by automation. True, such forecasts were made throughout the XX century, so that we can evaluate their accuracy and scope.

The calculations of pessimistic economists are binary: the profession will either disappear or not. But there is a middle ground. For example, AI threatens accountants. However, it can be assumed that only part of their activities will be automated.

Such forecasts take into account the technical probability of automation, but not profitability. From a technical point of view, the work of a bicycle repair wizard can be automated. From an economic point of view, this makes no sense: this is not the highest paid job, and the robot is not cheap.

History shows that, thanks to technology, more jobs were created than destroyed. Before the industrial revolution, many people worked in agriculture, but then many professions appeared in offices and factories.

In addition, it is necessary to distinguish between «closed» and «open» work. In closed (for example, window washers or drivers), the set of duties is strictly limited and they will be automated. Open work will only expand. For example, a chemist is an open work. Automation will only expand the horizons of science.

It is necessary to evaluate the element of humanity in such professions as a teacher, nurse, nurse. Would you like a robot to look after you in old age? This is one of those issues that we still have to consider.

Most likely, we will work less. Before the industrial revolution, many worked 60 hours a week, after which this figure was reduced to 40 hours, and annual leave was added.

And another prediction: the first victims of the digital revolution will be drivers.

Previously, machines replaced a person in physical labor, but now they can take on some mental work. What will we do? We will open people in ourselves again. Machines can learn to write music and novels, but we will appreciate the creations of people more - because they are based on human experience.

Perhaps we will rediscover crafts, the value of handicraft. Today, when applying for a job, STEM skills (science, technology, engineering and mathematics) are appreciated, but in 2062, our social and emotional intelligence will become the most important qualities .

End of war

One of the first to disappear is the profession of a soldier. Autonomous weapons systems (DIA) will be the third military revolution - after gunpowder and the atomic bomb. And the DIA is more dangerous than both of them. A small drone and a couple of grams of explosives do not require much scientific work, as is the case with nuclear weapons.

DIA will become an instrument of terror. Like any robots, they can be wrong, and the price of such errors is incredibly high. The author is convinced: we can control this process with the help of UN declarations and other decisions at a high level, although so far the situation is tight.

The end of our values

Machines do not share our values, but copy our prejudices. When translating from English into German, Google will translate «kindergarten teacher» in the female gender, and «teacher» in the male gender. The same Google for men offers more advertising for highly paid professions than for women. For Mac owners, sites offer more expensive hotels. Cars learn from biased data!

In 2015, Jackie Alsine discovered that Google Photos recognized Jackie and his girlfriend as a gorilla. The problem was not solved in a simple way, and Google simply removed the «gorilla» mark. The machine does not know what «racism» and «insult» are, and people neglect this knowledge. Facebook was selling ads aimed at «Jewish opponents.» He advertises his own vacancies mainly to young people and sells advertising that discriminates against the elderly.

Now the services tell us books, acquaintances, products ... The problem is that bias is not so easily eliminated. A large part of machine learning is deciding what kind of bias to give the program. The only way out is to turn bias into something more acceptable.

To entrust AI decision making, we need to clearly articulate our own values. In this sense, the «golden age of philosophy» is not the time of Aristotle and not the Confucian age, it begins now.

End of equality

We live in a happy time. For the first time, the number of people living in extreme poverty fell below 10%. Over 90% of the world's population under 25 can read. However, the gap between the lower strata and the rich has widened significantly, and the middle class is not involved in enrichment. By 2062, inequality in society will increase dramatically.

Welfare is concentrated not only in the hands of the rich, but also in the accounts of large corporations. The author goes through measures that will change the situation for the better: power over company directors, tax reforms, universal basic income ... These measures seem radical, but in reality this is only a continuation of what we are doing now (free education and medicine, etc.).

Fighting inequality will not be easy. It will require a better distribution of the benefits that AI will bring to large technology corporations.

End of privacy

In 2013, one of the creators of Google, Vint Cerf, said: "Private life may be abnormal," it is only a product of the industrial revolution. This is true: in the Middle Ages, for example, life was much less private. And yet today we feel close to the reality of 1984 .

They earn on our data. They say that data is new oil, only oil is an expensive and limited resource, while data is neither expensive nor scarce. And most importantly, oil is owned by states, and most of the data is private property. And that jeopardizes our privacy. However, Google and Facebook continue to grow rich on them.

In Silicon Valley, they know who you are, how you vote, what is your sexual orientation. The point is not only in the data itself, but also in their connectivity. The sender's address and the size of the message will not tell the attacker much; he does not see the contents of the message. But by combining them with other data, you can learn a lot about who you are talking to.

Even when we are offline, they still follow us. In 2013, it was discovered that «smart trash cans» in London were tracking mobile phones, and by 2062 the whole city will be watching us. Surveillance will not end even at your home: it’s not without reason that Zuckerberg and Snowden seal the camera on a laptop. Soon we will give way to our analog privacy - right down to the heartbeat that the fitness bracelet reads.

Our digital shells help us lie. But in the analog world it’s more difficult to lie - we cannot control the heartbeat. Imagine what a political party could do if it had access to data on our heartbeat. And we give such information to private companies.

The most alarming threat to our privacy is the social rating that is being developed in China. Europe is also aware, but it gives hope for privacy in the future. In May 2018, the General Data Protection Regulation (GDPR) entered into force. Its main goal is to let Europeans decide the fate of personal data themselves.

AI can protect our privacy. By 2062, smartphones will have enough power to autonomously support all processes: in order to recognize the voice of the owner, the smartphone will not need to contact Google. If we make the right choice, privacy will be a right secured by technology.

End of politics

There are already quite a few examples of how technology is changing the course of the political struggle: remember Egypt or Trump. At the same time, the greatest influence in social networks is not people, but cars. Trump has 14 million false pages out of 48 million followers on Twitter. The Pope has even worse performance: out of 17 million of its subscribers, 10 million are fakes.

By 2062, human voices will be lost among computer ones. When And And it becomes more reasonable, it will be more and more difficult for us to distinguish it from a person. A bot has already been created that successfully writes tweets like Trump.

We already know that photos and videos cannot be trusted. By 2062, it will be impossible to trust anything that we see or hear, if we are not present, we will not be able to distinguish a real politician from a false one. Do we want to find ourselves in a world where no more successful political ideas win, but better algorithms?

End of the west

Today, Google responds to 8 out of 10 search queries worldwide. But the Chinese search engine Baidu is in fourth place in the list of the most visited sites in the world. The American Amazon is valued at $ 750 billion, and the Chinese Alibaba at $ 500 billion, and it is developing faster.

China is stepping on its heels. The Chinese government plans to become the world leader in AI by 2030. There seems to be no reason to doubt success. China has lagged behind in this area for a long time, but is quickly adapting to new technologies: it is already a leader in the number of mobile payments. China has serious advantages in the race for AI: it does not bother with the privacy of its citizens.


the end

We are approaching a critical point in human history. We already have enough problems, and then there is AI. But the author is sure: there is no need to adapt to the future, it is the result of decisions in the present, so we can choose it.

We can consider laws restricting data ownership, and make platforms responsible for content. We need to separate large corporations in order to return competition to the digital market, and introduce new taxes so that corporations repay their debt to society. We will have to pass laws that prohibit political campaigning on social networks. Key words: we can and should.

We live much better than a hundred years ago, thanks to science. But not only she changed our life in the twentieth century. We have changed society to cope with the challenges created by progress. We came up with labor unions, labor laws, and universal education so everyone can benefit from progress. We must remember this at the beginning of a new technological revolution.