The Integrity of Psychology Assessments in the AI Age: A Critical Examination

Generative artificial intelligence’s (Gen-AI) increasing prevalence within higher education raises concerns over its usage by students, particularly over their development in skills like creativity and critical thinking. One important component of the standard undergraduate programme are assignments, where skill development can be assessed and constructive feedback offered to enable students to continue to build upon their skills. In this case study, we submitted the assignments from a three-year Psychology undergraduate programme to ChatGPT using three levels of prompting – basic, intermediate, and advanced – to answer the question, can Gen-AI obtain a psychology bachelor’s degree? Responses were reviewed by researchers experienced in assessment marking and deemed as pass or fail. We report on the finding that Gen-AI was capable of successfully passing almost all assessment types. Of 40 individual assessments, 36 were of passing quality. We discuss the implications of this susceptibility, where Gen-AI can be used positively as a learning tool but is simultaneously dangerous when it generates incorrect, but convincing, information. The positive marking approach enables incorrect values or basic formatting to be condoned reducing the desire or need for students to make significant effort to alter the Gen-AI content to evade detection. We recommend that traditional written assessments, where not invigilated, should be adapted either in terms of assignment or marking criteria to adapt to the aforementioned challenges.