Dean of teaching and learning Prof Tim Cooper said the university had been forced to respond to the technology’s growing influence and the answer did "not lie in policing and detection".
So, from next year Otago is planning to introduce a three-tier system for students.
AI use would be broadly broken down into three categories for assessments: restricted, guided and encouraged.
The change was included in a revision of the university's assessment policy which was presently under way.
It was necessary to approach the AI issue head-on.
"The amount of money that's been invested in AI is incredible, and the pace at which the technology is developing and expanding is eye-watering. So, it's a disrupter for universities.
"We need to think about what we offer to our students in the age of AI, and it is very much occupying our minds."
The university uses anti-plagiarism software such as Turnitin, which matches students' essays with an archive of material to pick up where students may have copied material in a way that is a breach of academic integrity.
Turnitin has an AI-detection feature.
"We have never turned that on, and we will not turn that on.
"The reason for that is that we do not want even one student to find themselves in a position of being wrongly suspected of improper use of AI because the tool is just not reliable," he said.
"The risk is too high that students will be falsely accused, and we do not want that for any student," Prof Cooper said.
Teaching staff were alert to the possibility that students were using AI, he said.
"The reality is that some students might be so proficient at the use of AI that you actually can't tell."
Taking a positive, proactive approach of rethinking assessment design in a way that encompassed AI and educated students in its use was the way forward.
"Guided learning takes place within an AI environment that we provide for the students, and we'll be providing it next year, for students and staff to use AI.
"It will be embedded within our learning management system, and that's where we will be allowing students to use AI in the course of their assessment, and we can see what they're doing."
Having visibility was key to the use of AI in an academic environment, he said.
"If you can see what they're doing, then it mitigates academic misconduct."
The university should not retreat into having more exams to halt AI’s presence.
"As a university, we're not going to simply retreat into hosting more exams. Instead, we are going to think in creative and effective ways, in research-informed ways, about assessment design because there are really, really effective assessment models that use AI. So again, it's encouraging the use of AI."
The structure of courses could change in the new environment, Prof Cooper said.
"There may well be, in contexts where the numbers allow this, more in the way of performance or oral interrogation of the material, other ways of being assured that the actual person is demonstrating the learning.
"But we would always want to do that in ways that are informed by good pedagogy and what's good for student learning."
