20Q
20Q is a computer based game that asks 20 questions. It was originally an artificial intelligence (AI) experiment .
The game
The game 20Q is a device based on the parlor game Twenty Questions . The device asks the player to think of anything and then tries to guess it by asking 20 questions that can be answered with yes or no.
Principle and history
20Q was created in 1988 as an artificial intelligence (AI) experiment . The principle is that the player thinks of something, the artificial intelligence asks a series of questions and guesses what the player is thinking. The artificial intelligence learns independently from the information it receives from the players and is not programmed. The player can answer the questions as follows: Yes , No , Unknown , or Sometimes . The experiment is based on the classic word game Twenty Questions and the computer game " Animals ", which was popular in the early 1970s and used a much simpler method to guess an animal.
The 20Q AI uses an artificial neural network to select and guess the questions. After the player has answered 20 questions (sometimes earlier), 20Q guesses. If it is wrong, it asks more questions and guesses again. It advises based solely on the information learned; it was not fed information or the inventor's opinion. All answers are based on the player's interpretations of previous questions.
The 20Q KI independently decides how to use the information. It could be described as a lay taxonomy rather than a taxonomy . Your knowledge grows with every game. In this regard, the online version of the 20 AI can be inaccurate as it relies on what people think rather than what they know . The boundaries of taxonomy are often overcome by the AI itself because it can learn and adapt. For example, if the player thinks of "horse" and answers no to the question "is it an animal", the AI will still guess correctly even though it has been told that a horse is not an animal.
Kevin Kelly, co-founder of Wired magazine, wrote:
- "Burned into its 8-bit chip is a neural net that has been learning for 17 years. Inventor Robin Burgener programmed a simple neural net on a DOS machine 1988. He taught it 20 questions about a cat. He then passed the program around to friends on a floppy and had them challenge the neural net with their yes / no answers to the object they had in mind. The neural net learns only when it plays a game; no data is added except for the yes / no answers of visitors. So the more people who test it, the more they teach it. In 1995 Burgener put the now robust neural net onto the new web where anyone could play it (that is, train it) 24 hours a day. And they did. Burgener's genius was to turn the hard tedious work of training a neural net into a fun game for humans. "
- "The 20Q AI is adaptable, scalable, modular and embeddable, and for this reason, it is possible for it to expand its knowledge and learn about more specific things. It is now learning in twenty-one languages, as well as everything it can about music, sports, movies and television. "
Translation:
- “Burned into its 8-bit chip is a neural network that has been learning for 17 years. The inventor Robin Burgener programmed a simple neural network on a DOS computer in 1988. He taught 20 questions about a category. He distributed the program to his friends on floppy disks and had them compete against the neural network with their yes / no answers to the object they were thinking of. The neural network only learns when a game is being played; no data is added to it except for the players' yes / no answers. So the more people have tried it, the more they teach it. In 1995 Burgener put the now robust neural network on the Internet, where anyone could play (read: train) with it around the clock. And they did. Burgener's talent was to turn the hard, tiring work of training a neural network into a game for people. "
- “The 20Q KI is adaptable, scalable, modular and concise and can therefore expand its knowledge and learn something about differentiated things. Today she learns in 21 languages as well as everything about music, sports, film and television. "
As described by the inventor Robin Burgener, the "unusual knowledge" that the AI spits out at the end of the game is what comes up when something seems strange and does not coincide with what it knows. This is what makes AI unique - it begins to make its own decisions; This is information that the AI itself comes up with by generating answers with what it has learned and what it knows. Over time, their knowledge will be refined even more. 20Q learns to make decisions by playing it - the more often an object is played, the more the AI learns about the object. If z. For example, if the player thinks of the object "particle accelerator", the game asks up to 20 target-oriented questions to adequately describe the object. The Online 20Q KI has around 10,000,000 connections. Burgener writes that the online 20Q KI's success rate is between 73 and 78 percent. According to Burgener, the real success rate could well be higher, but he adjusted the algorithm to make the game more interesting; if the AI won every game, as could theoretically be achieved in a closed system, the game would not be interesting and the AI would also not continue to learn.
If several players choose a certain object and the game asks different questions to different players, the large number of questions asked to the players will describe it in more detail. As a result, the game can guess the object "particle accelerator" faster and more accurately.
The entertaining factor lies in the fact that the player often chooses an object that he considers failed or unknown to the game in order to "outsmart" the game ("won't guess"). In fact, other players have often thought of the same object and so the game has extensive knowledge of the guessing object. This is often presented in such a way that the game asks seemingly diffuse questions where one thinks that they can hardly help the game in guessing the object and then suddenly guess exactly the right object.
The online version of 20Q is offered in 22 languages and the player can suggest his object to the system after the game if 20Q could not guess the object and does not yet know it. According to the website, the object database in the online version will be expanded to include the suggested terms after editorial control. Certain topics are censored for the protection of minors and the game refuses to continue guessing.
Artificial Intelligence modularity
The modularity of the 20Q KI means that it can be incorporated into devices with a small screen. At the moment there is a handheld version of the AI. The device contains a small fraction of the 20Q website database; In contrast to the online version, the handheld version cannot learn.
The 20Q KI differs from less flexible and extremely large expert systems. Its modularity, adaptability and scalability means that it can be transferred to other, more complex devices for more complex use.
- "Robin Burgener wants to turn child's play into rocket science. When he speaks to a room full of NASA scientists, programmers and technicians [in May of 2006], he'll explain how a simple parlor game he first adapted into a computer program 20 years ago might just be the answer to some of the agencies most pressing issues. "
- “Robin Burgener wants to transfer child's play to space research. When he speaks in front of a room full of NASA scientists, programmers and technicians (in May 2006), he will explain how a simple game, which he put into a computer program for the first time 20 years ago, is the answer to some of the most pressing problems the agency. "
See also
Web links
Individual evidence
- ↑ Information source: LiCalzi O'Connell, Pamela. "Vegetables And Minerals On The Radar," The New York Times. March 27, 2003; Burgener, Robin, computer architect, inventor.
- ↑ Kelly, Kevin. "Cool Tools." . March 21, 2005.
- ↑ Burgener, Robin. Computer architect. All figures come from data published by the inventor.
- ↑ 20q.net FAQ . Last accessed on March 24, 2018
- ↑ Harvey, Ian. "Is it animal, vegetable or artificial intelligence." The Globe and Mail , Toronto, Canada. May 2, 2006.