AI compatible assignments
2025 August 11 Monday 16:28
It has been of interest to investigate recently, how to use these various "artificial intelligence" tools (also exclusively of choice, the open source 'deepseek' service) Usage has prompted the proverbial question of how to devise assignments for students that will be resistant to completion with ai, in terms of minimal input.
As thoughts begin to turn (briefly) to the imminent new year, the following factors to evaluate assignments are proposed:
-
The assumption is that ai will be used to answer questions that comprise the assignment;
-
Citations must include the ai "prompt questions" applied, to the specified ai service;
-
Students will be told to expect random selection of "prompt questions" for review and potential discussion:
-
Are the ai results repeatable?
-
What was the rationale, the motivation for the "prompt questions" posed?
-
Have multiple ai services been used and cited; if not, why the reliance upon only one ai service?
-
How were the results from the ai service concluded to be true and appropriate to the assignment?
-
Evidence of ability to utilise other source that are not derived from ai, is expected
A question remains, how to evaluate a student that fails/refuses to declare use of ai, but proceeds to do so? Perhaps the default action is to scrutise with more rigour, those that opt not to use ai. In addition, as amgaf search engines begin to integrate more deeply these new ai services, it is anticipated that students may become less proficient to identify primary resources without ai services. This commercial integration may hinder development of explicit critical thought.
