This article discusses the concept of "Games With a Purpose" (GWAPs), which use computer games to solve computational problems and train AI algorithms. The authors, Luis von Ahn and Laura Dabbish, propose that instead of solely improving AI algorithms, humans can be used to perform tasks that computers cannot do, through the design of games that are enjoyable and engaging. The article outlines three general game templates for GWAPs: output-agreement games, inversion-problem games, and input-agreement games. These templates are used to create games that encourage players to solve computational problems as a side effect of playing. The goal is to make the games fun and engaging, so that players are motivated to play and contribute to the computational task. The ESP Game is an example of a GWAP where players label images as a side effect of playing the game. The game is fast-paced, enjoyable, and competitive, and has generated millions of labels for random images on the web. The labels can be used to improve web-based image search, which typically involves noisy information. Other GWAPs include Peekaboom, which locates objects within images, and Phetch, which annotates images with descriptive paragraphs. Verbosity is another GWAP that collects commonsense facts to train reasoning algorithms. The article also discusses the importance of ensuring output correctness in GWAPs. This can be achieved through various mechanisms such as random matching, player testing, repetition, and taboo outputs. These mechanisms help to prevent collusion and ensure that the data collected is accurate. The authors also discuss the importance of player enjoyment in GWAPs. They suggest that games should be designed to be fun and engaging, with features such as timed response, score keeping, player skill levels, high-score lists, and randomness. These features help to motivate players to play and contribute to the computational task. The article concludes by discussing the evaluation of GWAPs, including metrics such as throughput, lifetime play, and expected contribution. These metrics help to determine the effectiveness of GWAPs in solving computational problems and training AI algorithms. The authors also mention that further research is needed to improve the methods and metrics used to evaluate GWAPs.This article discusses the concept of "Games With a Purpose" (GWAPs), which use computer games to solve computational problems and train AI algorithms. The authors, Luis von Ahn and Laura Dabbish, propose that instead of solely improving AI algorithms, humans can be used to perform tasks that computers cannot do, through the design of games that are enjoyable and engaging. The article outlines three general game templates for GWAPs: output-agreement games, inversion-problem games, and input-agreement games. These templates are used to create games that encourage players to solve computational problems as a side effect of playing. The goal is to make the games fun and engaging, so that players are motivated to play and contribute to the computational task. The ESP Game is an example of a GWAP where players label images as a side effect of playing the game. The game is fast-paced, enjoyable, and competitive, and has generated millions of labels for random images on the web. The labels can be used to improve web-based image search, which typically involves noisy information. Other GWAPs include Peekaboom, which locates objects within images, and Phetch, which annotates images with descriptive paragraphs. Verbosity is another GWAP that collects commonsense facts to train reasoning algorithms. The article also discusses the importance of ensuring output correctness in GWAPs. This can be achieved through various mechanisms such as random matching, player testing, repetition, and taboo outputs. These mechanisms help to prevent collusion and ensure that the data collected is accurate. The authors also discuss the importance of player enjoyment in GWAPs. They suggest that games should be designed to be fun and engaging, with features such as timed response, score keeping, player skill levels, high-score lists, and randomness. These features help to motivate players to play and contribute to the computational task. The article concludes by discussing the evaluation of GWAPs, including metrics such as throughput, lifetime play, and expected contribution. These metrics help to determine the effectiveness of GWAPs in solving computational problems and training AI algorithms. The authors also mention that further research is needed to improve the methods and metrics used to evaluate GWAPs.