AI-Assisted Assessment and Bias in Algorithms

Statistical bias = Bias occurring due to the system being based on a biased dataset Algorithm bias = Bias introduced through poorly designed or implemented algorithms Human bias = Bias that happens when human reviewers are influenced by the results of an AI system Exclusion and marginalisation = Potential negative effects of AI if not implemented consciously

Tailoring lessons = Using AI to adapt lessons to each individual student Streamlining administrative processes = Using AI to make administrative tasks more efficient Insights into learner's habits and motivations = Using AI to gain a deeper understanding of each student's learning patterns

Potential for bias = Concern that algorithms may lead to unfair or inaccurate results Exacerbation of exclusion and marginalisation = Concern that AI may further isolate or discriminate against certain groups Faultline = An area of potential risk or conflict in the use of AI

Biased dataset = A cause of statistical bias where the data used to train the AI is not representative Poorly designed or implemented algorithms = A cause of algorithm bias where the algorithm itself contains or introduces bias Influence of AI system results = A cause of human bias where human reviewers are influenced by the AI's output

Unfair or inaccurate results = Possible outcome of bias in AI-assisted assessment Revolutionising the university system = Potential positive impact of AI use in higher education Exacerbation of exclusion and marginalisation = Potential negative impact if AI is not implemented consciously