Alex Ezat Parnia, President and CEO at Florida Coastal University
The academic integrity conversation taking place in universities concerning artificial intelligence is not an appropriate conversation, and the manner in which the discussion is underway is one piece of evidence. Universities are buying AI detection software, developing policies, and revising honour codes to include definitions of AI assistance and their consequences. Also, universities are convening assessment redesign teams to develop assessments that are either AI-proof or AI-resistant enough to be defensible tools for evaluating students. The above actions are definitely real and there is some necessity to these actions, but they are primarily revolving around the question that universities have sufficient time to answer and move beyond: how can we stop students from using AI? The much more relevant question is how can we create conditions for students to use AI in ways that support their continuing development instead of replacing it with AI?
It’s perfectly reasonable for a university to want to provide some security for its assessment process when a disruptive technology emerges. For example, the emergence of generative technologies that can produce applicant-level essays in less than 45 seconds creates an understandable response by universities to protect the integrity of assessment. However, detecting whether or not a student has used these technologies to generate their submission has become the dominant strategy rather than an emergency-response approach to protecting the integrity of institutions’ assessments. Additionally, universities are now permitting these detection-oriented tools used to respond to the emergency situation to become entrenched in the form of policy frameworks, which will ultimately shape how future professionals will work alongside AI systems, many of which will be far more capable than what caused the original panic.
Why Compliance Policies Are Teaching the Wrong Lesson
Particularly in environments where universities are still primarily focused on detection, as opposed to developing students’ capability to exercise judgment (i.e., knowing when it is appropriate to use AI and when it is not, but also, knowing whether or not to accept AI-generated output based upon how it has impacted their ability to think), students are not receiving the educational experience necessary to develop such judgment. Therefore, instead of relying solely upon compliance-oriented policies, universities must have an educational theory for using AI; this is significantly further behind the compliance side for many institutions.
Most faculty know that it is not the knowledge that a student gains, or even the credential received that creates value, but the intellectual changes that occur to them when forced to sit with a topic long enough to form their opinion. The essay, the report, the argument presented in class are not just ways of measuring a student's ability to do these things; they are providing an environment in which this development occurs. Using AI tools without Reflection violates the process of developing the student. The result is a person who receives their degree, but has not developed their thinking skills that the degree was designed to teach them, and who will not realize this until after they leave the safety of their learning environment and enter the working world where there will be no formatted tools from AI or others that allow them to compensate.
Teaching Students to Think About AI, Not Just Around It
That argument, made clearly and made to students rather than simply encoded into honour codes, is more powerful than any detection tool. It asks students to make a choice about their own development rather than to comply with a rule. Institutions where this conversation is happening openly with students through induction programmes, module introductions and feedback sessions are doing something that detection-first institutions are not doing – teaching their students to consider what they are doing with AI, rather than only whether they are allowed to do it.
The application of this methodology will provide students with a practical program that combines traditional and modern methods of education in order to develop their information literacy, with an emphasis placed upon teaching them to critically apply an analytical approach to evaluating all sources, including outputs from artificial intelligence. Students will be required to develop an analytical framework for evaluating both the output generated by the AI and any assumptions made by the AI based on its reasoning and the amount and type of evidence provided before concluding the accuracy of the information contained within the output. In conjunction with developing a critical analysis of AI-generated outputs, assessments will also need to be designed where the evaluation of AI-generated outputs will centre on the AI tool itself rather than the end user, and all students will need to understand that the learning process lies within the process of struggling with a difficult argument without the assistance of AI, and that this struggle is, therefore, not an inefficiency that should be eliminated or optimised.
The Professional Stakes of Getting This Wrong
Professional schools have a distinct duty here because the graduates of these schools will be going into professions that already have artificial intelligence incorporated into the existing practice, but this will only increase in the future. An attorney graduating from law school who has never been instructed on how to examine an AI-based legal research instrument to determine if the tool is revealing relevant data or concealing it, is unprepared for the practice of law in a responsible manner. A medical student who has utilized an AI support tool for diagnosing but has not had the opportunity to consider its limitations in terms of the data on which it was trained has a clinical judgement deficiency that a policy regarding such AI-based tools could never have detected.
Higher educational institutions are structured around the transfer of methods of thinking and not just the transfer of knowledge. The question that institutions must ask regarding their use of artificial intelligence is the same one that they have asked about the use of every other new tool incorporated into the educational process: how does using AI change how we think and how do we teach students to use AI in a manner that enhances their ability to think? Answering this question will help institutions understand that there are two very different questions related to AI's use, and they will need to pay more attention to this dichotomy.