Integrating AI into Educational Assessment
The use of artificial intelligence in assessment has become an integral piece of the discussion around integrating AI into education. Academic integrity is the main concern that is brought forward by faculty. However, many students would also express concern if their work was being marked by AI. Both sides of the educational equation do not wish to see important work offloaded onto artificial intelligence, but are there benefits?
Leaving an indelible mark on the landscape of tomorrow.
| Benefits of Integrating AI into Educational Assessment | Risks of Integrating AI into Educational Assessment |
| Marginalization and Inequity AI assessors are also prone to bias. If the information they are trained on has biases present this can lead to unfair outcomes. Students from under represented or marginalized groups could be impacted. Higher education is still predominantly attended by white students of the highest socioeconomic class. Further, faculty members and leadership within Ontario’s post secondary system are not well represented by minorities. These characteristics must be considered when developing AI Assessors. Data Privacy and Security If AI systems are given responsibility over the grading of assessments, this offloads significant data onto web-based systems. Historically, post-secondary institutions have been cautious to safe-guard student data, an easier task when stored on in-house systems. Privacy is not just a philosophy for these intuitions, there are government policies that enforce the implementation of robust data protection measures. The ability for these tools to safeguard student information is a major consideration during the investigation phase of implementing. Ethical and Transparency Issues Transparency and the ability to support grading metrics are imperative to providing fair assessments. AI systems are often much more opaque than a human instructor. It can be difficult for both instructors and students to trust the judgments made by AI without the ability to adequately discuss the nuances of how a decision was made. Furthermore, who is at fault if errors do occur? Do these issues fall on developers, educators, or administration? While thoughtful integration will involve collaboration between stakeholders, errors that impact student success create high-stakes scenarios for accountability. Scope and Reliability These tools still require human oversight, they are not immune to errors or breakdown. This creates new human resource challenges, and questions around who takes responsibility for these tools. Depending on the type of teaching and learning taking place AI tools may not be sufficiently equipped to do the assessment. Learning objectives focused on higher cognitive or creative domains of thinking may be too abstract or complex for AI models to evaluate effectively. | Personalized Learning Through AI assessment, new learning opportunities are presented to students. Once a student’s strengths and challenges can be identified, AI tools can create individualized learning plans and tailored instruction to improve student success. AI assessment is a complimentary piece to intelligent tutoring systems and can provide increased academic support to students without also increasing human tutoring resources. Applying the principles of Universal Design to Learning can be bolstered by these AI tools. Creating personalized tools for students based on their unique perspectives and characteristics ensures that the learning environment quickly adapts to their needs. Data Driven Insights One of the areas that AI Assessment tools excel is in making inferences about the assessments you have created. Do these assessments indicate overall success, do your evaluations scaffold well, and is there a particular student demographic that struggles in your course? These are all questions that can be answered quickly through AI assessment analytics. The insights gained from these analytics can support pre- and co-requisite decisions, help identify students who are struggling early, and determine if an assessment needs to be redesigned or replaced. AI tools can quickly analyze large pools of data, across multiple semesters, allowing you to make better curricular decisions. Efficiency and Automation Especially for formative testing, feedback needs to be timely. Utilizing AI grading tools for formative assessments allows students to have a clear understanding of the areas the need to work on throughout their studies. Being able to provide student feedback, in some cases in real-time, ensures that they can come prepared for face-to-face lectures or activities. Automating this process also allows for the faculty to focus on patterns rather than specifics. Recognizing that the majority of the class is struggling with the same topic, quickly, allows for adjustments to be made more efficiently. Adaptive Testing Computer adaptive testing allows for students to be challenged at an appropriate level throughout their evaluation. This type of testing makes sure that learners are evaluated at a pace that reduces anxiety and better evaluates capability. |