The History of AI
A quick summary in the history of technology.
We have come a long way in technology since our ancestors first invented the abacus calculator. Machines have continued to improve thanks to the hard work of the generations before us. Nevertheless, we find ourselves at a time where there is allot of improvement to be made. This is both scary and exciting because we will all have the opportunity to contribute to the growth of technology, either directly or indirectly. However, before we are able to make an informed decision on this topic we need to learn the of how AI works and the rate at which technology is improving by looking at two central characteristics of intelligence; problem solving, and language.
Problem Solving
AI was first coined back by computer scientist John McCarthy in 1955 [1]. However, its earliest age can be traced back to the 1950s when American computer scientist Arthur Samuel programmed a machine that could play checkers [2]. Machines playing video games are a good way to track the progress of AI because it requires problem solving skills, which is a good way to measure intelligence [2]. This machine used heuristics to play checkers meaning it had many basic rules of thumb programmed for what would be considered a good move, or a bad one. The program was simple but revolutionary for its time. The next computer gaming machine to revolutionize the industry came when IBM developed Deep Blue [1]. Deep Blue was programmed to play chess, and unlike its predecessor who used heuristics, Deep Blue used brute force to play chess. This meant it calculated every possible move that could be made and chose the one that gave it the highest possibility of winning [2]. This eventually allowed Deep Blue to beat Garry Kasparov, the world chess champion to this day [1]. Most recently a computer program by the name of AlphaGo was programmed to play the board game GO, a sort of Chinese chess game [1]. AlphaGo reverted to using heuristics to play the game since the game is famous for having more playable moves than there are atoms in the universe [2]. The program worked by adding layers of different categories for what would be considered a good or bad move and choosing the one that best suitifies each layer. For example, one layer would be looking to only include legal moves. Another layer would be looking to find an uncontrolled area and so on [2]. It also analyzed the board to see if it was something it had seen in the past [2]. In 2017 AlphaGo beat world champion Ke Jie [1]. This was seen as the final test for an AI with problem solving skill, meaning there are not many ways left that machine learning can improve in this area [2].
Language
One of the characteristics that is unique to humans is that we can communicate with Language. This is something no other species that we know of can do. So when Joseph Weizenbaum pioneered a chatbot that could speak to people named Eliza in 1964 it was seen as a pretty big deal [1]. Weizenbaum created Eliza by making it look for key words in a sentence and looking at the order of these words to guess what you were trying to say [3]. This method proved effective and is still used to this day. This idea was revolutionary and would not be dramatically improved until apple released Siri in 2011 [1]. Siri was revolutionary in its own way because it integrated a voice interface into the device that functioned as a virtual assistant [1]. It then looks to see if it is a problem that can be easily solved like telling it to “play my music”, or a more complex problem like “what is the temperature in Michigan” [4]. If it is an easy problem it will solve it itself, if it’s a hard problem it will go onto the apple servers to solve the problem [4]. Another advancement in language happened this year when two Facebook chat-bots created their own language when speaking to each other [5]. Facebook quickly pulled the plug, reportedly because it was not the task they had set out to accomplish [5]. This is big news because it proves that machines are able to create their own information, not just respond the way they are programmed.
Artificial Intelligence Timeline
Artificial Intelligence Today
[1] Marsden, Paul. “Artificial Intelligence Timeline Infographic – From Eliza to Tay and beyond.” Digital Intelligence Today, Digital Intelligence Today, 21 Aug. 2017, digitalintelligencetoday.com/artificial-intelligence-timeline-infographic-from-eliza-to-tay-and-beyond/.
[2] SciShow. YouTube, YouTube, 14 Sept. 2016, www.youtube.com/watch?v=Xhec39dVGDE.
[3] YouTube, YouTube, 6 Apr. 2017, www.youtube.com/watch?v=EUWlXtrRlMY.
[4] MrYouMath. YouTube, YouTube, 20 Dec. 2011, www.youtube.com/watch?v=loOHmMFVJcE.
[5] CBSNewsOnline. “Facebook scraps A.I. chatbots after they created their own language.” YouTube, YouTube, 1 Aug. 2017, www.youtube.com/watch?v=ONPqeHJShdQ.