Shulou(Shulou.com)06/03 Report--

Https://www.toutiao.com/a6662955854104363532/

Alan Turing, the father of computer and artificial intelligence, asked: can machines think? If a machine can talk to humans without being identified, then the machine has intelligent characteristics. At the same time, Turing also predicted the possibility of creating machines with true intelligence.

The birth of AI

At a conference held by Dartmouth College in 1956, artificial intelligence was formally established as a research discipline.

Parties reunited at Dartmouth meeting in 2006, from left: Trenchard More, John McCarthy, Marvin Minsky, Oliver Selfridge, Ray Solomonoff

The best part of the first development (1955-1974)

The Dartmouth Conference was followed by an age of great discovery. For many people, the program developed at this stage is magical: computers can solve algebra problems, prove geometric theorems, learn and use English. Among the many studies, search reasoning, natural language and micro-world were the most influential at that time.

Achievements at that time

Artificial neural network was proposed in the 1930s and 1950s, and the first neural network machine was made by Marvin Minsky in 1951.

Behrman formula (rudiment of reinforcement learning) is proposed.

-Perceptron (prototype of deep learning) is proposed

-search reasoning is proposed

-Natural language has been proposed

-it is proposed for the first time that artificial intelligence has the characteristics of imitating intelligence, knowing how to use language, forming abstract concepts and solving existing human problems.

The chess program developed by Arthur Samuel in the mid-1950s and early 1960s has been able to challenge amateurs at a considerable level.

The Robot SHAKEY project is highly publicized and can "reason" about its behavior; it is regarded as the first general-purpose robot in the world.

-the proposal of the micro-world

The first cold winter-- existing problems

Researchers believe that fully intelligent machines will appear within 20 years and give a lot of predictions, such as machines will be able to do all the work that adults can do, and will build a machine with average human intelligence.

But it soon hit me in the face, and AI encountered a bottleneck. As expectations were too high because of previous over-optimism, when AI researchers failed to deliver on their promises, the public began to criticize AI researchers, and many institutions continued to reduce funding for artificial intelligence research until funding was stopped.

-the computing power of the computer has encountered a bottleneck and cannot solve the complex computing problem of exponential explosion.

Common sense and reasoning require a lot of information about the world, and computers cannot "understand" or "understand".

-unable to solve the Moravik paradox

-unable to solve some logical problems involving automatic planning

-Neural network researchers are left out in the cold

Moravik Paradox: if a machine plays chess like a mathematical genius, how hard is it that it can imitate a baby's learning? However, this has proved to be quite difficult.

The best part of the second development (1980-1987)

The AI program of "expert system" has been adopted by companies all over the world, and artificial intelligence research has ushered in the best part of a new round. Because the "expert system" is limited to a very small field, the problem of common sense is avoided. "knowledge processing" has also become the focus of mainstream AI research.

Achievements at that time

-the birth of expert system

-AI researchers have found that intelligence may need to be based on multiple processing methods for a large amount of classified knowledge

-BP algorithm has achieved a breakthrough in neural network training, and neural network researchers have attracted more and more attention.

AI researchers proposed for the first time that in order to achieve true intelligence, a machine must have a body, which needs the ability to perceive, move, survive, and interact with the world. Perceptual motor skills are very important for high-level skills such as common sense reasoning, and the reasoning ability based on things is more important than abstract ability, which also promotes the development of natural language and machine vision in the future.

The existing problem-the first cold winter

The market demand for AI hardware has suddenly dropped, and although the expert system is very useful, its application field is too narrow, and the cost of update iteration and maintenance is very high.

The main problems at that time were:

-impacted by the concept of desktops and "personal computers"

-the pursuit and snub of AI by commercial organizations turned AI into a bubble and burst.

-the bottleneck of computer performance still can't be broken.

-there is still a lack of massive data training machines.

The best part of the third development (1993-so far)

Under Moore's Law, computer performance continues to make breakthroughs. With the rapid development of cloud computing, big data, machine learning, natural language and machine vision, artificial intelligence ushered in the third best part.

Achievements at that time

1997

-IBM's chess robot Deep Blue defeats world chess champion Kasparov

2006

-Geoffrey Hinton proposed a deep learning algorithm for multilayer neural networks.

-Eric Schmidt put forward the concept of "cloud computing" at the search engine conference

2010

-Sebastian Thrun-led Google driverless car was exposed, setting an accident-free record of more than 160,000 kilometers

2013

-Deep learning algorithm has made a breakthrough in speech and visual recognition rate.

2016:

-Google AlphaGo beat nine-stage go player Lee se-dol at 4:1

The concept of Chatbots is becoming popular.

-Google releases the first generation of dedicated chip TPU customized for machine learning

-Google publishes voice assistant Assistant

The above is the general history of the development of artificial intelligence, quickly forward it, Amway to your partner!

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.