By Dr. Ramakrishnan Raman, Vice Chancellor, Symbiosis International (Deemed University).
Rethinking the Purpose of Assessment
The introduction of artificial intelligence (AI) technologies in universities is changing the way students study, solve problems and prepare for professional life. New subjects related to AI and generative AI in education are becoming popular in universities. However, while focusing on development of AI courses, the problem of assessing student's performance is often being ignored. For several decades, universities used examinations, assignments, presentations and dissertations as assessment tools. Currently, AI systems can produce essays, analyze cases, write programs and prepare business reports within minutes. As a result, the quality of the submitted task may not reflect the knowledge of a student, his or her understanding of material and effort put in the preparation process.
From Evaluating Outputs to Evaluating Understanding
Current forms of assessment tend to focus more on the output of the student than on the way he or she reached a particular conclusion. In the era of AI tools, current assessment practice requires changes. For instance, management students can use AI for developing a comprehensive marketing entry strategy including competitor analysis, segmentation of the customers and financial forecasts. But the student does not necessarily understand all assumptions that were made during the process. Therefore, the task of assessment should include explanation of strategic decisions made by the student, validation of data used, identification of weak spots and defence of suggested strategy in case of changes. Similarly, law students should prove that they can interpret legal principles and precedents, and criticize AI-generated arguments.
Redesigning Assessment for an AI-Enabled World
It does not mean that universities should discard the traditional form of assessment. Instead, a combination of conventional and new methods can be implemented. Oral examination, viva voce and discussions can help university instructors to make sure that students understand their work. Assessment based on the process may require the explanation of the research process conducted by a student, identification of the help that was provided by AI and of places where the student decided to accept or reject AI suggestions. Real world projects involving businesses, communities or professionals can assess students' skills of applying knowledge in practice. Open resource examinations can ask students to analyze AI-produced work, find mistakes and offer corrections. For example, finance students can receive a financial forecast produced with the help of AI and identify its errors and propose solutions for making the work better.
Academic Integrity and Responsible AI Use
Growing popularity of generative AI systems raises concerns about plagiarism, improper use of AI and academic dishonesty. However, treating any use of AI as improper would be impractical and educationally wrong. Instead, universities should provide clear guidelines of situations when AI is prohibited, allowed or recommended. Students should know whether it is possible to use AI for brainstorming, improving language, writing code, analyzing data or producing the whole piece of work, and if he or she needs to declare that fact. It is also important that universities do not rely solely on software that detects the use of AI since those programs are unable to prove authorship of a student's work and may give incorrect results. Assessment of academic integrity issues should take into account multiple factors including the ability of a student to explain his or her work and conditions of its creation. Honest use of AI requires transparency, verification, accountability and respect for academic regulations.
Preparing Graduates for the Future Workplace
Changing the approach to assessment is also necessary for preparing students for working in an environment dominated by AI technologies. Companies will look for employees who can analyze information, question assumptions, solve problems and make right decisions. As a result, critical thinking, AI literacy, communication, ethics and adaptability should be the key learning outcomes across disciplines. Business students can use AI for creating a business strategy and then defend it in front of a board of managers. Engineering students can analyze AI-generated design of a building or a system against various criteria like safety, cost and sustainability. Health sciences students can analyze recommendations generated by AI considering evidence and professional and patient safety. It is assumed that graduates should be aware of limitations of AI, verify the output of algorithms and make decisions independently when automated suggestions are not applicable.
The Role of Faculty and University Leadership
The introduction of transformational changes in assessment requires action at an institutional level and cannot be left to individual academics. University administration needs to start with the analysis of current assessment practices to find out the tasks where the output of AI will be satisfactory without demonstrating the required learning outcomes. It is necessary to train faculty staff to design authentic assessments, perform efficient oral evaluations and assess works done with the help of artificial intelligence. The access to technologies needs to be guaranteed by universities; this becomes even more important when it comes to assessments requiring AI tools from students. The differences in access to paid subscriptions, hardware and internet need to be compensated to avoid discrimination. Universities need to test innovative assessment tools, receive feedback from students and faculty and analyze the improvements in learning outcomes. Assessment frameworks have to stay flexible since the proper ratio of tests, projects, oral evaluations and work with AI will depend on disciplines and programs.
A Practical Way Forward
The university could start by auditing the current assessments, introducing clear AI use policies and diversifying evaluation techniques. Writing assessments could be combined with oral defense, problem-solving under supervision, practical projects and reflection exercises. AI literacy needs to be included into all disciplines in such a way that the students learn how to validate outputs, know about the limitations of this tool and use it responsibly. Above all, it is crucial for universities to measure whether the students are able to explain, apply, question and defend their conclusions and not only focus on the output quality. These changes can be implemented gradually and universities would have a possibility to test their effectiveness.
Assessment Must Evolve with Education
In conclusion, artificial intelligence makes it necessary for universities to rethink the ways of assessing student's learning outcomes. The solution is neither prohibiting the usage of AI totally nor permitting its use without control. Universities need to create conditions for using this tool in the process of studying that will allow students to use the power of AI while retaining the independence of thinking and evaluating arguments and being responsible for the decisions.
