BLOG

THE ORIGINS OF ARTIFICIAL INTELLIGENCE

Did you know that Artificial Intelligence (AI) can be traced back to the beginning of the 20th century? While it is true that only science-fiction writers talked about it back then, Alan Turing thought that since we, as humans, can use the information that's available to us to take intelligent decisions, machines could do the same.

The young mathematician and computer scientist didn't have access to today's modern computers, of course. And the existing computing units were unable to store the received information; they could only process it, and then output the results of their execution.

Fortunately, things changed for the better within the next decade. Computers started to incorporate memory units, which made it possible to store the data, process it, and then store it again. Their prices fell dramatically, making the purchase possible for wealthy individuals and medium-sized companies.

ELIZA, a natural language processing application which was created at the MIT AI lab in the 60s, used artificial intelligence to simulate conversations with humans. ELIZA processed the users' input, and then used predefined scripts to respond with non-directional questions.

The program was a success, so the Defense Advanced Research Projects Agency (DARPA) decided to fund several AI projects; they wanted to get access to a machine that could transcribe and translate spoken language automatically. It was an ambitious idea, but it was doomed to fail, because computers weren't fast enough back then.

The first fully functional deep learning algorithm was published by Alexey Ivakhnenko in 1965. Then, John Hopfield and David Rumelhart popularized this essential subset of machine learning methods.

During the 80s, Japan decided to invest 400 million dollars into the "Fifth Generation Computer Project" program. The goal was to build computers which are very easy to use and have elementary reasoning abilities, being able to perform complex tasks on their own.

The Fifth Generation was supposed to understand and work with symbols and concepts, something that even modern computers haven't managed to tackle successfully yet. So, the program had a moderate success; most international scientists agreed that while the computers were innovative, they weren't revolutionary.

World champion Gary Kasparov was defeated by IBM’s Deep Blue chess computer for the first time back in 1997. The computer relied mainly on brute force computing power, evaluating up to 200 million positions per second. This allowed it to search for, and then test potential moves up to eight steps ahead.

Deep Blue learned from 700,000 chess games played by various grandmasters, and its victory proved that the company has made significant progress in the AI field. And while Kasparov thought that human chess players had intervened on behalf of the machine, IBM stated that human intervention occurred only between games.

In October 2015, DeepMind's AlphaGo has secured the victory against the world's number one Go player Ke Jie. The world's top Go playing computer utilizes a Monte Carlo-based search algorithm, trying to discover the best next move by exploring its database, which has been built by analyzing lots of human and computer games.

So, what will be the future of artificial intelligence? These days we already have driverless cars, and many scientists believe that AI-powered machines will soon surpass human intelligence.