AI streamer with feedback to AI streamer based on spectators
A method is provided, including: executing a session of a video game; executing an artificial intelligence (AI) player that performs gameplay in the session of the video game; streaming video of the AI player's gameplay over a network to one or more spectator devices for viewing by one or more spectators respectively associated to the one or more spectator devices; receiving, over the network from the one or more spectator devices, feedback data indicating reactions of the one or more spectators to the video of the AI player's gameplay; adjusting the gameplay by the AI player based on the feedback data.
1 . A method, comprising:
executing a session of a video game;
executing an artificial intelligence (AI) player that performs gameplay in the session of the video game;
streaming video of the gameplay by the AI player over a network to one or more spectator devices for viewing by one or more spectators respectively associated to the one or more spectator devices, each of the one or more spectator devices having at least one recording device;
receiving, over the network from a respective spectator device, audio data captured by the respective spectator device, wherein the audio data corresponds to speech of a spectator associated to the respective spectator device; and
adjusting the gameplay by the AI player by at least:
identifying positive feedback and negative feedback based on the audio data;
associating the positive feedback and the negative feedback with respective gameplay actions performed by the AI player;
adjusting one or more parameters of a ML model associated with the AI player; and
controlling the execution of the AI player in the session of the video game to increase a performance by the AI player of gameplay actions associated with the positive feedback, and decrease the performance by the AI player of gameplay actions associated with the negative feedback.
2 . The method of claim 1 , wherein adjusting the gameplay by the AI player based on the audio data includes, analyzing the audio data to determine a level of interest of the one or more spectators.
3 . The method of claim 2 , wherein adjusting the gameplay by the AI player is configured to increase the level of interest of the one or more spectators.
4 . The method of claim 1 , wherein the audio data is recorded using a microphone associated with the respective spectator device.
5 . The method of claim 1 , wherein the method is performed in substantial real-time, such that the adjusting of the gameplay by the AI player is responsive to the audio data during the session of the video game.
6 . The method of claim 1 , wherein the ML model is further trained based on a human player's gameplay.
7 . A non-transitory computer-readable medium having program instructions embodied thereon that, when executed by at least one server computer, cause said at least one server computer to perform a method including the following operations:
executing a session of a video game;
executing an artificial intelligence (AI) player that performs gameplay in the session of the video game;
streaming video of the gameplay by the AI player over a network to one or more spectator devices for viewing by one or more spectators respectively associated to the one or more spectator devices, each of the one or more spectator devices having at least one recording device;
receiving, over the network from a respective spectator device, audio data captured by the respective spectator device, wherein the audio data corresponds to speech of a spectator associated to the respective spectator device; and
adjusting the gameplay by the AI player by at least
identifying positive feedback and negative feedback based on the audio data;
associating the positive feedback and the negative feedback with respective gameplay actions performed by the AI player;
adjusting one or more parameters of a ML model associated with the AI player; and
controlling the execution of the AI player in the session of the video game to increase a performance by the AI player of gameplay actions associated with the positive feedback, and decrease the performance by the AI player of gameplay actions associated with the negative feedback.
8 . The non-transitory computer-readable medium of claim 7 , wherein adjusting the gameplay by the AI player based on the audio data includes, analyzing the audio data to determine a level of interest of the one or more spectators.
9 . The non-transitory computer-readable medium of claim 8 , wherein adjusting the gameplay by the AI player is configured to increase the level of interest of the one or more spectators.
10 . The non-transitory computer-readable medium of claim 7 , wherein the audio data is recorded using a microphone associated with the respective spectator device.
11 . The non-transitory computer-readable medium of claim 7 , wherein the method is performed in substantial real-time, such that the adjusting of the gameplay by the AI player is responsive to the audio data during the session of the video game.
12 . The non-transitory computer-readable medium of claim 7 , wherein the ML model is further trained based on a human player's gameplay.