Max Tegmark is a Swedish-American physicist, a professor at the Massachusetts Institute of Technology and a cosmologist. In his book "Life 3.0. Being Human in the Age of Artificial Intelligence" Tegmark explores the possibilities of artificial intelligence development, notes options for the future of mankind under the control of AI and analyses why we need artificial intelligence in general.

The subject of the book is complex and has a depth of presentation close to the scientific work. The following important points, in my view, are worth sharing:

• No one really knows, when and if (general) artificial intelligence (AI) will be developed. And even if it is invented in the near future (decades), no one will switch it on it until security issues are resolved.
• AI is good at achieving its goals, so we need to align its goals with ours. Example: you don't kill ants just like that. But if you are instructed to build a hydroelectric power plant and in an area that is flooded, there will be an anthill, the ants will have to say goodbye to life. So the movement for a favorable AI wants to prevent a situation, when humanity will be in the place of ants.
• The main AI danger is if its goals do not coincide with ours.
• The method of trial and error is not suitable for AI. We must act ahead of time for the security of mankind.
• In the long run, using your biometric data, algorithms will be able to create personalized melodies just for you. Or using an array of biometric data from millions of people, the algorithms will create world hits.
• AI will replace people in many professions, but also create new jobs - for remote management, AI improvement, data analysis and cybersecurity. Most likely, by 2050 the labor market will look like a collaboration between man and AI. Yes, human and AI teams will outperform human and computer teams. AI can also help train people - bankers, chess players, soldiers, detectives.
• New jobs in the AI era will require new knowledge and skills, which will not solve the problem of incompetent employees. It is one thing to switch from one low-skilled job to another: a land worker - a handyman in a factory - a cashier in a supermarket.
• - Most likely, we will immediately suffer from high unemployment (those who will not be able to retrain), shortage of qualified personnel (due to new jobs), and a whole class of people, who will not be able to adjust to the new reality, in particular, due to mental instability.
• Creation of new jobs and retraining of people will take place constantly. A successful industrial revolution also requires revolutions in teaching and psychology.
• Two scenarios for the safety of low-skilled staff: the government can pay people for work in a broad sense - helping neighbors, volunteering, caring for children. And the second option - the government can subsidize free education, medicine, transportation and more. AI in finance is the accuracy of forecasts, in production it is less routine and more safety, on the roads it is less deaths due to accidents, in energy it is reasonable energy consumption, in health care it is more accurate diagnoses and accuracy of surgical operations, in legal sphere it is no prejudices and prejudices.
• The higher is the level of automation of society and the more developed is AI, the more damage will be caused by cyber warfare. If you hack access codes and hack the programs of transport, drones, nuclear reactors, production robots, communications and financial systems, along with power grids, you can destroy any economy and paralyse any country and its defenses.
• Career guidance for children - choose professions that are not suitable for machines, require interaction with people, have elements of unpredictability and creativity.
• To make the Terminator scenario that will captivate humanity realistic, you need: Step 1. Create a human-level AI. Step 2. Use AI to create superintelligence. Step 3. Allow superintelligence to take over the world.
• Imagine totalitarianism in the AI era: the state monitors and "understands" all electronic communications (mail, SMS), listens to telephone conversations, has access to all surveillance cameras, analyzes all transactions with payment cards, with data from mobile operators understands where someone is, through electronic watches and phones writes voice, health data, movement. It remains to put on forced electronic bracelets on people with the ability to inject paralyzing substances and totalitarianism will be in action. So even the police will be no longer needed.
• People and machines can merge together by improving biological bodies (cyborgs), or by loading human consciousness into machines. These ideas do not violate any known laws of physics, and therefore can be implemented. A matter of time.
• Why would AI kill people? Take, for example, elephants - because they too could think: "Why should people kill us? And they are so small". But people are killing them with bullets, inventing technology, poisoning drinking water, destroying wildlife and transporting elephant tusks around the world to make different things out of them. So, if a clever AI wants to kill us, he can handle it quickly.
• Without technology, humans are threatened with rapid extinction (in the cosmic context of tens of billions of years), which will turn our entire history of existence into a short burst of beauty. With technology, we will increase the chances of both winning (gaining the potential to survive) and losing.
• If we do not improve our technology, the question will not be whether humanity will die, but only how it will die.
• A superficial AI will be able to easily achieve its goals, and if these goals are not agreed with ours, then we will be in trouble.
• Aligning our goals with AI goals is very difficult. There are the following issues: how to make AI understand our goals, accept our goals and stick to our goals.
• It is absolutely unclear how to set the ultimate AI goal, which will be clear, unambiguous and will not lead to the destruction of humanity.
• Neurobiological experiments show that much of our conscious experience is the result of subconscious processing of an immense amount of information.