Controlling generative language models for artificial intelligence characters

A computing platform for providing an Artificial Intelligence (AI) character model, the computing platform comprising:

• a processor;
• and
• a memory storing instructions that, when executed by the processor, configure the computing platform to:
• determine, based on a data stream received from a client-side computing device associated with a user, a context of a dialog between the AI character model and the user in a virtual environment, the AI character model and the virtual environment being displayed on the client-side computing device by an application running on the client-side computing device, the application communicating with the computing platform using a pre-determined communication protocol associated with the application;
• receive a message of the user in the dialog;
• pass the context and the message through a plurality of heuristics machine learning models to produce intermediate outputs of the plurality of heuristics machine learning models;
• upon the producing the intermediate outputs of the plurality of heuristics machine learning models, compose, based on the intermediate outputs of the plurality of heuristics machine learning models and a predetermined templated format ingestible by a language model, dialogue prompts being an input to the language model configured to predict a response to the message, the context including parameters of an event occurring in the virtual environment independently of the dialog between the AI character model and the user;
• upon the composing the dialogue prompts, provide the input to the language model to obtain the response, wherein the language model is configured to:
• process the input to form a request for a large language model, the processing including at least one of classifying and adjusting the message based on the context;
• provide the request to the large language model;
• receive, from the large language model, a model response to the request;
• and
• modify the model response from the large language model based on predetermined criteria to form the response, wherein the modification includes adding, to the model response, a specific word corresponding to a current emotional state of the AI character model;
• and
• transmit the response to the client-side computing device, wherein the client-side computing device presents the response to the user.

A non-transitory computer-readable storage medium, the computer-readable storage medium including instructions that, when executed by a processor of a computing platform providing an Artificial Intelligence (AI) character model, cause the computing platform to:

• determine, based on a data stream received from a client-side computing device associated with a user, a context of a dialog between the AI character model and the user in a virtual environment, the AI character model and the virtual environment being displayed on the client-side computing device by an application running on the client-side computing device, the application communicating with the computing platform using a pre-determined communication protocol associated with the application;
• receive a message of the user in the dialog;
• pass the context and the message through a plurality of heuristics machine learning models to produce intermediate outputs of the plurality of heuristics machine learning models;
• upon the producing the intermediate outputs of the plurality of heuristics machine learning models, compose, based on the intermediate outputs of the plurality of heuristics machine learning models and a predetermined templated format ingestible by a language model, dialogue prompts being an input to the language model configured to predict a response to the message, the context including parameters of an event occurring in the virtual environment independently of the dialog between the AI character model and the user;
• upon the composing the dialogue prompts, provide the input to the language model to obtain the response, wherein the language model is configured to:
• process the input to form a request for a large language model, the processing including at least one of classifying and adjusting the message based on the context;
• provide the request to the large language model;
• receive, from the large language model, a model response to the request;
• and
• modify the model response from the large language model based on predetermined criteria to form the response, wherein the modification includes adding, to the model response, a specific word corresponding to a current emotional state of the AI character model;
• and
• transmit the response to the client-side computing device, wherein the client-side computing device presents the response to the user.