The Impact of Artificial Intelligence on University Assessments: A Wicked Problem
Artificial intelligence (AI) is fundamentally disrupting traditional university assessments and examinations. With generative AI tools like ChatGPT enabling students to create essays and complete assessments in a matter of seconds, the academic landscape is experiencing significant upheaval. Universities are in a race against time, forced to redesign evaluation methods, update academic integrity policies, and integrate new detection systems to counteract this growing trend of academic dishonesty.
The Challenge of Academic Integrity in the AI Era
The integration of AI into educational frameworks is not merely a challenge for institutions; it places immense pressure on educators and the validity of degree programs. Current students need assessments that provide real value, while academic communities and employers seek assurance that university qualifications maintain their worth. The dilemma becomes evident: how can we uphold the integrity of assessments in an age dominated by AI?
Recent research suggests that the complexity of the AI-assessment problem is far more challenging than current media narratives indicate. It is not simply a matter of finding one correct solution; rather, the situation embodies characteristics of a "wicked problem."
Understanding Wicked Problems
Coined by theorists Horst Rittel and Melvin Webber in the 1970s, the term "wicked problem" refers to issues that defy straightforward solutions. Examples include climate change, urban planning, and healthcare reforms. Unlike "tame" problems—those that can be resolved with sufficient time and resources—wicked problems offer no single, definitive answer. Instead, they present a spectrum of better or worse options.
Wicked problems are inherently messy, interconnected, and resistant to closure. Any attempt to "solve" these issues tends to introduce new complications, trade-offs, and unintended consequences. Recognizing that there are no straightforward solutions allows for an appreciation of the complexities involved and acknowledges the necessary compromises that must be made.
More Read
Insights from Recent Research
In a recent study, we engaged with 20 university educators from various faculties at a large Australian university to understand the impact of generative AI on assessment design. These interviews provided enlightening perspectives—revealing that many educators view the challenge posed by AI as fundamentally dreadful and characterized by unavoidable trade-offs.
One instructor shared:
"We can make assessments more AI-proof, but if we make them too rigid, we just test compliance rather than creativity."
This sentiment captures the essence of the dilemma: the goal is not merely to eliminate AI’s influence but to achieve a balance that promotes genuine learning.
Another educator posed a critical question:
"Have I struck the right balance? I don’t know."
Such reflections underscore the uncertainties educators face as they strive to create assessments that reflect both real-world capabilities and academic integrity.
Navigating the Trade-offs
Educators grapple with various considerations, such as whether to allow AI in assessments. Should they embrace it, recognizing that students will encounter AI tools in professional environments? Conversely, should they exclude AI to test independent capability? Similarly, oral examinations have been posited as a form of assessment less susceptible to AI interference, but they also place extra burdens on educators and may disadvantage some student demographics.
The logistical challenges are daunting. One instructor remarked:
"250 students by […] 10 minutes […] it’s like 2,500 minutes. That’s how many days of work to administer just one assessment!"
Similarly, while in-person, hand-written exams present their own challenges, they fail to assess the full range of skills required in modern workplaces. Balancing multiple assessments throughout a course further complicates matters.
The Dynamic Nature of AI
As AI technology continues to evolve, educators face an additional layer of complexity. Generative AI tools, including ChatGPT, are constantly releasing new models and functionalities. This relentless pace of change means that educators must not only revamp their assessment strategies but do so in alignment with evolving technological capabilities.
Yet, amidst all these challenges, educators are expected to manage their usual teaching responsibilities, which are already demanding. This combination leads to a chaotic situation that lacks closure. One educator succinctly summed it up:
"We just do not have the resources to be able to detect everything and write up any breaches."
A New Approach to Assessment Design
Recognizing that the AI-assessment dilemma is a continuously evolving wicked problem is crucial. Holding onto the notion that it can be neatly "solved" has profound implications—not only for educators but for policy-making and institutional frameworks. Trying to impose quick fixes can lead to burnout, anxiety, and a constant cycle of policy churn.
Instead of pursuing absolute solutions, universities should foster environments that encourage continuous negotiation around assessment practices in the context of AI. This can alleviate the pressure on educators, allowing them to focus on creating adaptable, transparent processes that acknowledge the inherent trade-offs involved.
Our research highlights the need for universities to grant educators greater "permissions" to explore flexible, context-specific approaches to assessment. This might entail compromising on certain ideals to better cater to their unique academic environments, students, and curriculum demands.
Acknowledging the intricate trade-offs involved in our assessments can pave the way to developing practices that support diverse learning pathways while maintaining academic integrity. The challenge remains formidable, but understanding the nuance of these decisions can foster a more resilient educational landscape.
Inspired by: Source
