The vast majority of students rely on laptops – and increasingly AI – to help with their university work. But a small number are going analogue and eschewing tech almost entirely in a bid to re-engage their brains
The vast majority of students rely on laptops – and increasingly AI – to help with their university work. But a small number are going analogue and eschewing tech almost entirely in a bid to re-engage their brains
o.o holy shit- i mean that’s a valid move, using AI for a handwritten piece sounds like a pain in the ass, but so does just writing 10 pages by hand, AI or not!
i’m glad i got through my higher education marginally before the AI boom hit (i graduated 3 years ago). i only had Turnitin yell “PLAGIARISM???” at me when i used a common phrase that another student used at some point somewhere (think - “The research suggests…”, or sometimes even the page numbers), good times good times
LLM boom has certainly affected education - complicating things for honest students and at the same time empowering cheaters.
Having studied both pre- and post-boom, I can say the amount of times I was offered to use LLMs overall and ChatGPT/Gemini specifically to generate answers as a student has gone through the roof.
And as a soon-to-be educator (I currently pursue PhD and aspire to teach others), I collect ideas on how to combat it, as it tanks the quality of education so much it may as well be nonexistent. But in any case, students that genuinely complete their assignments should not be harshly affected.
my best idea would be going old school with in person written & oral testing, since clearly nothing digital is of any help anymore. or perhaps require multiple digital WIP versions to be submitted? would also be getting the students into a good habit of making backups of their work. or maybe every essay should come with a director’s commentary (a more loose style reflective essay on the research and work done)
These are all good options! In person testing is certainly on my list, and I like the ideas with WIP versions (especially for larger submissions) and commentary.
I also think of more presentation format submissions where I could ask quick questions to see if the person actually understands what is written. Sort of a small defense.
On technical means, I welcome different forms of AI poisoning in tasks: these don’t always work, but they can catch the least attentive.