Human upgrade: can we think as fast as machines

There is no doubt that computers are now not only able to solve simple computational problems faster than humans

In some strategy or even creative games, it can also beIn support of the latter, they usually citecases when a computer, playing chess, the game of Go or the intellectual quiz "Geopardi", defeated the best minds on the planet.Amazing precision and colossal memory size are really the strengths of the computer, and in Geopardi you can also note what we humans call creativity.

However, comparing the work of the human brain withwith such algorithms is incorrect. It is many times more complex and capable of outstanding results. The fact that we are inferior in this battle to the artificial algorithm speaks of some vulnerabilities, but not of its characteristics. A person can still come up with a new algorithm for solving what a computer is not capable of. IBM Watson, of course, is no longer a calculator, but it is not at all a person with the ability to think and solve complex problems with synthesized semantics. For humans, as a representative of a living species, the process of consuming information and issuing a decision over billions of years of evolution has developed into a complex nervous system capable of finding solutions to problems, not only synthesized from various sources of information, but in general in the absence of such information. Let's try to figure out how this happens.

IBM Watson is an IBM supercomputer equipped withAI system and created by a group of researchers led by David Ferucci. Its creation is part of the DeepQA project. Watson's main job is to understand questions formulated in natural language and find answers to them using AI. Named after IBM's first president, Thomas Watson.

Neurointerface

Any information that comes to a personfrom the outside, processed by the brain, and the output is a signal. In this system, all the senses - sight, smell, hearing, taste and touch - represent an almost infinite number of combinations of various information, but by processing this flow, the brain can respond only with nerve impulses. The movement of the eyes, hands and body is entirely dependent on nerve impulses directed from the brain to the muscle tissue. This limitation raises a logical question for researchers: is it possible to spy on a person's thoughts by extracting these nerve impulses directly from the brain?

At the moment, in medicine and in many high-tech businesses, special interest and hopes are pinned on neural interface systems.The NS is a kind of intermediary in the interaction of the brain with the external system without direct contact.In science fiction literature, this would be called thought control or telekinesis — or even the transfer of human consciousness into a computer.The huge interest in this area led to the appearance of the popular series "Black Mirror" or the cult film "Ghost in the Shell".The narrative "World of the West" or the quantum DEVS are all works of fiction that speak to the same scientific hypotheses and discoveries.

This area of development has great potential: from explicit everyday tasks, such as turning on the television with the power of the mind, to controlling complex systems or even controlling artificial organs.In 2016, Nathan Copeland was able not only to shake President Obama's hand, but also to feel it through sensors.

How a machine reads thoughts

The human brain is part of a huge communication network, the central nervous system.This, in turn, expresses activity in the continuous creation of electrical potentials in the nerve cell, redistributing positive ions and creating depolarization, a potential difference.Depolarization accumulates over time and the negatively charged neuron becomes positively charged, it is "pierced", and the impulse is transmitted further.This process was inspired in 1943 by Pitty McCulloch, who created the first artificial neural network.

Electrical activity reading technologyThe nervous system is not innovative and has already given birth to such important medical techniques as the electroencephalogram. Recording brain activity during various motivations applied to the brain cannot simply be counted and used directly; each brain is unique in this sense. It is also worth noting that the brain is an ever-changing architecture that is constantly recording experience and memory and growing neural connections. Therefore, the interface uses a large amount of training data to ensure that the algorithm is stable against inevitable changes.

Registration and recording are lengthy processes,requiring a lot of effort and time, until recently even dangerous, but not basic. The main task of the neurointerface is to decode the received information and receive a response under the conditions that the data is a multidimensional space with a high degree of noise. At this stage, artificial neural networks play an important role. Despite being primitive compared to the real nervous system, they can find structure in spaces that are difficult for humans to imagine. Moreover, at the moment the neural network can decompose the brain activity and adapt to a specific operator. To unravel the language of communication of brain cells with each other, we need a system that vaguely resembles the work of the brain. This is a real paradox.

Help neurointerface

As stated earlier, the brain and featureshis activities - the architecture is fickle, changing, and it varies greatly from person to person. What can we do to make our approaches universal?

Neurobiology is a science that is not alien to patterns,can we solve them and use them to make life easier for the neural interface? Our observation of the psyche through the prism of various sciences from mathematics to biology has led to ideas that very clearly describe the real state of affairs. These ideas formed the basis of the theory of thinking formats.

Just as we feed data into a neural network, so does ourthe brain receives information. Just as a neural network must find some image from the chaos of numbers or pixels, so our brain must make a decision from a continuous flow of information from different senses. How this systematization occurs and what decision is made at the output determines the type of network or type of thinking.

Despite the clear analogy with a neural network,the theory itself is not simple in nature and requires for analysis a number of characteristics that need to be measured - speed, structure and volume of thinking. In this context, the analogy of water flow from hydrodynamic physics is applicable: flows are viscous, laminar and stable, or turbulent. Theoretically, knowing the input parameters, we can roughly describe the behavior of the flow in the long term and make predictions for some future, which, surprisingly, will lie within a narrow confidence interval.

We do not pursue the goal of describing everything to the smallest detail, thisimpossible, but we assume that there is some basic system of patterns that guides the personality. These are postulates of truth for a person, which change slightly over time. The principle that the theory deduces is quite consistent with the logic of the technology. I think many people know the phenomenon of "survivor error". During World War II, the Hungarian mathematician Abraham Wald had to find a solution to an important problem: which parts of the bombers should be strengthened. The planes returned to the base with holes and even lost parts, but the engines were intact. The mathematician made a paradoxical conclusion: you need to strengthen those parts that have survived. He realized that the parts that fell off were not so important, since the bombers were able to return, but what remained intact is the most important, and without these parts the planes would not have been back at the base. Likewise, people who are trying to develop in themselves what "fell off" by nature are most likely mistaken, because they survive precisely due to their unique "engine".

In this paradigm we see opportunities to determinea format of thinking for using the obtained data in a more structural interaction with the neural interface and overcoming the vulnerability of great uncertainty.

Elon Musk recently announced that his companyNeuralink, which develops neurointerfaces, will begin testing its products on humans as early as next year. The entrepreneur is already thinking about a device with which people could communicate without words with their devices. Elon estimates that it will take only 5-10 years to develop the technology that will make this possible. At the moment, the speed of communication between our smartphone and the brain is very low: you need to look at the screen, make a series of touches to the sensor, or dictate something to the phone orally. As soon as the neurointerface as a mediator is connected to the brain, it will be instantaneous - at the level of thinking. And this is where the fun begins. Because here a person has just more chances to achieve the very speed of the solution, which in narrow directions is available only to modern machines. It remains only to aggregate artificial neural networks and natural ones. But that's another story.

See also:

Look at the huge “wall” of hundreds of thousands of galaxies behind the Milky Way

Comet NEOWISE is visible in Russia. Where to see her, where to look and how to take a photo

It turned out that made the Mayan civilization leave their cities