Realtime facial biometric feedback for dynamic optimization of artificial intelligence output

A computer program product comprising a computer readable storage medium having a computer readable program stored therein, wherein the computer readable program, when executed on a computing device, causes the computing device to perform real-time modification of an interaction with an artificial intelligence (AI) computer model based on detected real-time passive responses from a user to content generated by the AI computer model, at least by:

• recording by one or more sensors, responsive to the user consuming an output by the AI computer model as the output by the AI computer model is generated in real-time, a dynamic passive user response of the user;
• analyzing, by a machine learning trained computer model, the recorded dynamic passive user response in real time to classify the recorded dynamic passive user response with regard to a plurality of predetermined classifications corresponding to different levels of comprehension by the user or user emotional states;
• assigning, by the machine learning trained computer model, user scores corresponding to the plurality of predetermined classifications to a portion of the output by the AI computer model;
• responsive to at least one user score of the assigned user scores exceeding a threshold score, pausing generation of the output by the AI computer model;
• in response to pausing the generation of the output by the AI computer model, generating a prompt specific to the portion of the output and a level of comprehension classification or an emotional state classification of the user corresponding to the at least one user score;
• and
• resuming generating the output by the AI computer model, from the portion of the output, in response to the AI computer model processing the generated prompt.

An apparatus comprising:

• at least one processor;
• and
• at least one memory coupled to the at least one processor, wherein the at least one memory comprises instructions which, when executed by the at least one processor, cause the at least one processor to perform real-time modification of an interaction with an artificial intelligence (AI) computer model based on detected real-time passive responses from a user to content generated by the AI computer model, at least by:
• recording by one or more sensors, responsive to the user consuming an output from the AI computer model as it is generated in real-time, a dynamic passive user response of the user;
• analyzing, by a machine learning trained computer model, the recorded dynamic passive user response in real time to classify the recorded dynamic passive user response with regard to a plurality of predetermined classifications corresponding to different levels of comprehension by the user or user emotional states;
• assigning, by the machine learning trained computer model, user scores corresponding to the plurality of predetermined classifications to a portion of the output;
• responsive to at least one user score of the assigned user scores exceeding a threshold score, pausing generation of the output;
• in response to pausing the generation of the output, generating a prompt specific to the portion of the output and a level of comprehension classification or an emotional state classification of the user corresponding to the at least one user score;
• and
• resuming generating the output, from the portion of the output, in response to the AI computer model processing the generated prompt.