Artificial intelligence: the biggest lie of the century that nobody can recognize

As the concept of “artificial intelligence” to fool people and why even the most educated investors do not have the power to recognize this trick? About it writes RBC.

Искусственный интеллект: самая большая ложь века, которую никто не может распознать

Photo: Depositphotos

In March 2018 the McDonald’s Corporation acquired a startup, Dynamic Yield over $300 million the fast food giant decided to use machine learning technology to personalize the client experience.

In the age of artificial intelligence (AI), this solution looks reasonable and even obvious, the more Dynamic Yield widely known for its AI development, and even got into a prestigious ranking of the best startups in this field. So it might be considered that the neural “maxeta of” already set to work.

But there is one problem. According to the publication, the former Director of content Dynamic Yield”s Mike Millazzo portal Medium, the startup has no relation to artificial intelligence. Later, the author removed his full of sorrow exposing artificial intelligence, but the collective memory of the Internet is preserved.

Milazzo said that marketers, investors, experts, journalists and programmers involved in the fraud with AI technology. The very definition of artificial intelligence, he writes, “is so complicated that any use of that term at first glance seems justified.”

Criticism of Malacca is based on two things. First — intentionally-deceptive marketing gimmicks that often come into play in any high-tech sectors. Here it is sufficient to recall the history of the little-known Long Island Iced Tea, whose shares soared by 289% after it changed its name to Long Blockchain, hinting at the possibility of the development of blockchain technology.

The second problem is that, unlike the blockchain, the concept of artificial intelligence is very vague, which allows the use of the term in any situation when it comes to technology for the solution of difficult tasks. Moreover, this problem arose long before the AI was at the peak of popularity. To better understand when it all started, go deeper into the story of the creation of this technology.

The development of artificial intelligence emerged as an independent science in 1956 during a summer workshop at Dartmouth College. From the description of the seminar showed that in just two months, participants will be able to greatly succeed in getting machines to understand language, to generalize and form concepts, solve problems, cope with that is only people, and improve themselves.

The goal, announced by the creators of the workshop, was so striking that they still are a source of inspiration for modern scientists. I must admit that since then it’s been much more than two months, and mankind is no closer to solving tasks, but the ideas of the Dartmouth seminar gave impetus to the larger studies.

But the successes to date achieved by scientists, hardly even reach a vague definition of the AI that made the pioneer in this field, Marvin Minsky (he defined AI as the science of creating machines capable of performing tasks that require the use of intelligence in the case of fulfillment of their human).

For example, consider a subsection of AI as heuristic search, the development of which began in the 1960s. Then, the Stanford scientists wanted to create a robot that can move independently, avoiding obstacles in its path. Continuing the nomenclature trend in science, the researchers have dubbed their creation the Neck, and the first algorithm to determine a path called A1. It was succeeded by the no less remarkable algorithm A2, later renamed to A*.

As it turned out, the principles of movement from one point to another, and puzzle solving are very similar. So A* can be considered a multifunctional algorithm, and scientists believe that it is one of the most powerful tools in the Arsenal AI. However, this algorithm is so simple (he decides what action to take next, just adding two numbers) that can hardly be an alternative to human intellect.

A similar story could be told about other developments in the field of AI, among which multi-agent systems (focused on building interaction between Autonomous software like, for example, is used in unmanned vehicles), automated planning, and machine learning, which many mistakenly think is synonymous with artificial intelligence. The main achievements in these areas have not yet even reach theses Minsk.

All these seemingly different areas are United not only by shared history and excessive optimism. As in other fields of science in the field of AI have in common terminology, this means that the most promising ideas and powerful technologies in General are a kind of advertising of other subsections of artificial intelligence.

In addition, at the intersection of different fields of AI from time to time appear extremely ambitious undertaking. For example, in the 2000s, there were competitions DARPA Grand Challenge and DARPA Urban Challenge, which gave impetus to the development of unmanned vehicles.

In 2011, the IBM supercomputer Watson crushed the two Champions of the game show Jeopardy. Also recently a lot of studies United by a common goal to direct AI technology to solve the main problems of humanity.

The moral is that the term “artificial intelligence” really is often misused. But the concept originally included a variety of technologies. So that investors and journalists should demand from Treponema on the use of AI startups explanations of what they are doing.

The Council is not to judge a book by its cover, maybe syassen, but it is doubly fair to the era of artificial intelligence. And even more fair, if we are talking about a book written by artificial intelligence.