here is an idea, curious what others think of this:
split the entire coursework into two parts:
part 1 - students are prohibited from using AI. Have the exams be on physical papers than digital ones requiring use of laptop/computer. I know this adds burden on corrections and evaluations of these answers, but I think this provides a raw answer to someone's understanding of concepts being taught in the class.
part2 - students are allowed, and even encouraged to use LLMs. And they are evaluated based on the overall quality of the answer, keeping in mind that a non zero portion of this was generated using an LLM. Here the credit should be given to the factual correctness of the answer (and if the student is capable of verifying the LLM output).
Have the final grade be some form of weighted average of a student's scores in these 2 parts.
note: This is a raw thought that just occurred to me while reading this thread, and I have not had the chance to ruminate on it.
split the entire coursework into two parts:
part 1 - students are prohibited from using AI. Have the exams be on physical papers than digital ones requiring use of laptop/computer. I know this adds burden on corrections and evaluations of these answers, but I think this provides a raw answer to someone's understanding of concepts being taught in the class.
part2 - students are allowed, and even encouraged to use LLMs. And they are evaluated based on the overall quality of the answer, keeping in mind that a non zero portion of this was generated using an LLM. Here the credit should be given to the factual correctness of the answer (and if the student is capable of verifying the LLM output).
Have the final grade be some form of weighted average of a student's scores in these 2 parts.
note: This is a raw thought that just occurred to me while reading this thread, and I have not had the chance to ruminate on it.