The Artifice

Generative AI’s biggest strength is also its greatest flaw.

As part of the process of creating human-like responses, it makes it look too effortless, seamless, like it’s all original ideas completely from scratch. In this case AI’s development model goes against the grain of what we’re looking for as teachers, as the artifice of ingenuity is crucial to its magic, but the black box it creates around its process is decidedly “unacademic”. Can we leverage these AI tools so that we’re using the tech rather than being used by the tech? Can we wrangle academic rigour from these tools of the Wild Wild West? Let’s find out:

Auditing for Complexity

The easiest exam questions to write are also the ones that don’t test for higher-order learning. 

Define this term… 

List 5 things about this topic… 

These are information retrieval tasks rather than evidence of deep cognition, and I’ll give you one guess what is simply amazing at retrieving information?

Googling your exam question used to be the only way to do it, and you still had to click across a few different sites to see the whole picture.  If the answer comes up straight away - probably something that won’t work in open-book exams, and really something that doesn’t test for higher-order learning like apply, compare and contrast, or justify and evaluate.  AI takes this a step further - it synthesises information from multiple sources, (none of which you’re supposed to know about!), to come up with a holistic answer for your question. 

How long is this answer? 

How repeatable is it if you ask AI the same question again and again? 

Does it stretch itself and describe things you didn’t even consider when you wrote the question? 

If it’s short, repeatable, and predictable, even with all the ingenuity and algorithmic might of AI, then your exam question likely falls short of the mark.

We’re teaching because we’re the expert right? 

We know all the right answers and the students don’t, so naturally marking should be this ephemeral process, our gut feeling on how “good” any piece of student work is.  Sadly this no longer passes the muster with teaching quality assurance agencies and assessment standards frameworks, and students have a right to see a “model answer” to a question and where their response falls short.  I don’t know about you, but these model answers take me ages to write every semester.  I’m also adapting them after reading through a few student responses, because usually they’ve considered something I haven’t - it’s a drawn out iterative process that takes much longer than it should. 

So why not use AI to do the first pass through, and use AI’’s response as the skeletal starting point of the model response you want students to come up with?  Don’t take the AI generated response at face value, but compare it to student responses, add to it, and amend it - it’ll still save you time overall.

Our working days are all filled with 100 tiny repetitive tasks, all of which take less than a minute to do but make you lose focus on the big picture.  Set an out of office message in Outlook.  Collate files in a specific folder based on their modified date.  Upload a YouTube video.  All of these are possible to do using APIs and basic scripts, but what if you don’t have any coding experience?  AI to the rescue - just ask it to write a script, and it’ll come up with a template for you to fill in the essentials yourself - file names, local directories - but the heavy lifting will be done for you.

There’s no quicker way to look foolish online than to be a so-called “expert” making bold predictions about the future of tech.  If you’re right half the time, you’re also wrong half the time, and that’s really the best case scenario for most pundits in this space.  The coverage of AI that I’ve seen is overwhelmingly negative and gloom and doom, and while I think that’s perfectly valid the only way to move forward is to engage with the threat in small but meaningful ways.  I want to integrate AI into my teaching to acknowledge the broader world around us to our students. 

If AI can “replace” how teachers assess students, what else can it replace?  Will the jobs they go into after graduation still look the same in not 10 years, but in 3 years’ time?  The onus is on all of us to learn how to use the technology rather than be used by it.

Jack.