Collecting Evidence
The process is no different than a “regular” academic integrity case. Members have been gathering “tangible” evidence (as you have with that screen shot), but they have also taken the time to meet with the student and asking them for not only clarification, but also for them to provide their notes, mind maps, graphic organizers (anything that demonstrates their writing process) because those are indicators of their thought process.
Also, they suggested using “old school ideas” and getting students to provide you with the actual resources they used. If you cannot find the resources on Google Scholar, then something is wrong. Another interesting piece of information (and this is coming from faculty members that have used Chat GPT)….this technology is designed to have a “conversation” with you. This means, if you ask for further clarification about resources, ChatGPT can’t do that, and if it does – it usually fabricates something. Which is another means of accumulating evidence, if you can get Chat GPT to provide you with a reference in the student’s assignment, and it doesn’t exist – that’s cheating according to the policy. They admitted that plagiarism evidence may be difficult to find, but cheating evidence is easier (and applicable).
One faculty member mentioned how they continued to ask ChatGPT follow-up questions about resources and it would fabricate the resources and create “a salad” of references, which were no where to be found. It looks like the “fabrication” avenue is the best option.
Plagiarism Checkers
An overwhelming number of faculty members admitted to these technologies having shortcomings, and one of those is their unreliability in providing you with accurate and usable “evidence.” Also, if you run a portion of the student’s work through a plagiarism checker, and it comes back perfect – that’s a red flag. Despite an individual writing their own assignment, the likelihood of there being no plagiarism is highly unlikely due to the scarcity of “original” thought. Oddly enough, they’ve run some of their writing through a plagiarism checker, and it’s come back as “some plagiarism detected.” Faculty members mentioned how this may show up because (as you know) the writer makes comments related to the evidence, and those thoughts have been likely mentioned by someone else. They all agree that these checkers are problematic.
Language Proficiency
Some members mentioned using their previous assignments as a benchmark toward assessing their language and writing proficiency. As you know, if a student has dramatically increased their comprehension in a short amount of time – that’s a red flag. Keep in mind – the AIC has deemed “text spinner technology” such as Grammerly an appropriate tool to use because it edits their work and doesn’t write the work for them. I’ve used it and it functions the same way as a tutor would – it offers suggestions on editing. Also, some faculties at TRU have partnerships with NVIT and instructors at NVIT have approved Grammerly as an appropriate application to use. Grammerly is not evidence, but…if you meet with a student and ask them for their notes (to which you have noticed a vast improvement) and they don’t mention Grammerly or another text spinner technology, then that’s evidence.
Collusion with Contract Cheaters
One member mentioned how Thomas Lancaster (who’s been doing research on this for some time and is a great resource) has uncovered contract cheaters (such as CHEGG) are now using ChatGPT. This adds even more complexity to the issue. And…. Microsoft and Google have added ChatGPT as well. In
response to this, some faculty members noted how some of their colleagues tried posting “deceptive assessments” to CHEGG and other contract cheaters to lure students into cheating. All these cases were dismissed because they trapped students. This would be the same outcome at TRU.
Members then went on to mention changing the assessment types and gravitating away from the formal essay can be one way. Here is an online link about what this member mentioned. https://teachonline.ca/ai-resources/tools-trends/when-chatgpt-makes-it-easier-cheat