training a conversational agent to simulate a job interview, wherein the conversational agent is trained by a regression model using historical interview dialogs and a corresponding job posting with labeled fit scores for one or more answers;

generating, by the trained conversational agent, a system output, wherein the system output is a question directed to a user interacting with the conversational agent;

receiving, by the trained conversational agent, a user input from the user interacting with the conversational agent, wherein the user input includes a response to the system output;

determining, by the conversational agent, that a system output/user input pair is not satisfactory based, at least in part, on the response to the system output included in the user input having a score below a predetermined threshold; and

generating, by the conversational agent, and in response to determining the system output/user input pair is not satisfactory, a counterfactual simulation of the response included in the user input based, at least in part, on a target intent of the response included in the user input, wherein generating the counterfactual simulation of the user input includes generating a counterfactual text modifying the user's response until the score is above the predetermined threshold.