GAME PLAN DEMYSTIFIED Decoding the logic behind two player games with AI
Game playing Games have been considered one of the foremost demonstrations of artificial intelligence. Why do AI researchers need to study game playing? It is a good reasoning problem, formal and non trivial. Direct comparisons with humans and computers is possible
What kind of games? Sequence of moves to play. Rules that specify possible moves. Rules that specify a payment (utility function) for each move. Objective that maximizes that payment .
Benchmarks in game development Chess and Deep Blue: IBM's Deep Blue, a supercomputer, made history in 1997 by defeating the reigning world chess champion, Garry Kasparov. Black & White: "Black & White" (2001) by Lionhead Studios introduced a notable use of machine learning to train creatures based on player behavior. Go : Games like "AlphaGo" (2016) showcased the capabilities of AI in strategic decision-making, influencing how AI could be integrated into strategy and board games.
The Turks "The Turk," a famous 18th-century automaton that played chess. The Turk was created by Wolfgang von Kempelen, and it toured Europe and the Americas from 1770 to 1854. The interesting aspect was that it appeared to be a mechanical chess-playing machine, but it was actually operated by a human chess master hidden inside the apparatus. Amazon Mechanical Turk (MTurk) is an online crowdsourcing marketplace launched by Amazon in 2005. It connects businesses and individuals (known as requesters) with a distributed workforce of workers, often referred to as "Turkers."
Difficulties in using search algorithms Unpredictable opponent specifying a move for every possible move by the opponent. Time limits unlikely to find a goal, find an approximate solution
Adversarial search It involves the systematic exploration of possible moves and counter-moves to make strategic decisions that optimize the chances of winning or achieving a desirable outcome. T he Minimax algorithm, evaluation functions, and alpha-beta pruning techniques contribute to the effectiveness of this approach.
CUTTING OFF SEARCH Depth Limiting: Setting a maximum depth for the search tree means that the algorithm will only explore a certain number of moves ahead. Beyond this depth, the search is cut off, and an evaluation function is used to estimate the desirability of the resulting positions. Time Limiting: Instead of limiting the depth, the search is cut off after a certain amount of computational time has passed. This approach allows for a more flexible exploration of the game tree, considering as many moves as possible within the given time frame.
Terminologies in adversarial search: States: -board configurations Initial state: -the board position and which player will move. Successor function: -returns a list of (move, state) pairs, each indicating a legal move and the resulting state. Terminal test: -determines when the game is over. Utility function.
Games solved by computers When machines think…
CHECKERS CHESS Chinook ended 40 year reign of human world champion Marion Tinsley in 1994. Used a precomputed endgame database defining perfect play for all positions involving 8 or fewer pieces on the board, a total of 444 billion positions. Checkers is now solved. Deep blue defeated human world champion Garry Kasparov in a six game match in 1997. Deep Blue searches 200 million positions per second, uses very sophisticated evaluation, and undisclosed methods for extending some lines of search upto 40 ply. Current programs are even better, if less historic.
TIC-TAC-TOE GO Tied for the best player in the world. 2016, Deep mind’s Alphago defeated Lee Sedol the world champion to end the human reign. Earlier, human champions refused to even compete against computers which were too bad. Alphago was developed after much research after applying successive pruning.
"In the realm of games and in the journey of life, every decision, every choice, is a move that shapes your destiny. Whether on a board or in reality, play wisely and embrace the adventure."