FWIW I teach online asynchronous and the AI battle is real. I am of course unable to prove most of it but have designed higher-level assignments so AI can't pass them without out a lot of input (requires a paid subscription) and multiple sophisticated prompts. As I teach middle-and working-class freshmen who can't afford and don't know how to work the better systems, this results in massive numbers of Ds and Fs. It does not promote honesty, alas.
The conversation we're starting to have in my department is whether we should even continue to offer asynchronous courses, if we cannot provide reasonable assurances that the students and not AIs are the ones completing the coursework -- and to what extent can we possibly provide those assurances. It's a big deal.
I have also gone to only in-person assessments for my engineering courses. My struggle is that I can’t assess more than 2 learning targets in a 50 minute class. So new attempts tend to be out of class and assessment dates seem too frequent. Any suggestions welcome!
I would take a look at those learning targets and see if maybe they can be broken down into smaller units, so there would be more of them but each one takes less time. In my classes, each Learning Target is set up to be assessable in 5 minutes or less if a person has prepped well for them.
Not sure if I agree with everything you wrote, but yes, the “everything is an in-class test” is the only sensible way to go. The challenge now is to redefine what “test” means in that context. Design disciplines have been using this approach for a long time. And your observation about anxiety is also spot on. Neurodivergent students are tough to integrate in such an educational approach. Thanks for sharing your thoughts!
Yeah, there is no way my school would ever eliminate asynchronous online classes now. And so far they are offering no guidance on how to combat AI cheating.
FWIW I teach online asynchronous and the AI battle is real. I am of course unable to prove most of it but have designed higher-level assignments so AI can't pass them without out a lot of input (requires a paid subscription) and multiple sophisticated prompts. As I teach middle-and working-class freshmen who can't afford and don't know how to work the better systems, this results in massive numbers of Ds and Fs. It does not promote honesty, alas.
The conversation we're starting to have in my department is whether we should even continue to offer asynchronous courses, if we cannot provide reasonable assurances that the students and not AIs are the ones completing the coursework -- and to what extent can we possibly provide those assurances. It's a big deal.
I have also gone to only in-person assessments for my engineering courses. My struggle is that I can’t assess more than 2 learning targets in a 50 minute class. So new attempts tend to be out of class and assessment dates seem too frequent. Any suggestions welcome!
I would take a look at those learning targets and see if maybe they can be broken down into smaller units, so there would be more of them but each one takes less time. In my classes, each Learning Target is set up to be assessable in 5 minutes or less if a person has prepped well for them.
Not sure if I agree with everything you wrote, but yes, the “everything is an in-class test” is the only sensible way to go. The challenge now is to redefine what “test” means in that context. Design disciplines have been using this approach for a long time. And your observation about anxiety is also spot on. Neurodivergent students are tough to integrate in such an educational approach. Thanks for sharing your thoughts!
My school's all in. You can earn your entire degree online now.
I love teaching online but do not believe that should be a thing.
Yeah, there is no way my school would ever eliminate asynchronous online classes now. And so far they are offering no guidance on how to combat AI cheating.