Investigating the affordances of OpenAI's large language model in developing listening assessments