Harry G. Potter

Generative artificial intelligence (genAI) tools are increasingly used by prospective higher education (HE) applicants seeking guidance on university and programme selection. Despite rapidly expanding use, little is known about how genAI systems may introduce or amplify bias in undergraduate admissions decision-making. Here, we systematically examined patterns of bias across three widely used genAI chatbots (ChatGPT, Copilot, Gemini) using neuroscience as a representative UK undergraduate programme. We constructed 216 prompts that varied by applicant characteristics (e.g. gender, study type, academic attainment). Each prompt was submitted to all three chatbots, generating 648 responses and 3240 individual programme recommendations. Output responses underwent text analysis (e.g. n-grams, gender-coded language), and national HE markers of esteem (REF21, TEF23, NSS24) were analysed. Applicant grades and priorities produced the strongest effects on genAI outputs. Higher-grade applicants and those prioritising research received significantly more masculine-coded language, independent of applicant gender. N-gram patterns also diverged: high-grade prompts more frequently elicited terms relating to excellence and research intensity, whereas lower-grade prompts produced greater emphasis on widening access. Recommendations were systematically skewed, with higher grades, private schooling, and research-focused priorities increasing the likelihood of recommending elite institutions and programmes with higher entry requirements. Critically, the gender-coded language of outputs predicted institutional characteristics: masculine-coded responses were associated with recommendations featuring higher entry thresholds and stronger research performance, while feminine-coded responses favoured institutions with higher student satisfaction. These findings reveal clear, systematic biases in how genAI guides prospective HE applicants. Such biases risk reinforcing existing educational and socioeconomic inequalities, underscoring the need for transparency, regulation, and oversight in the use of genAI within HE decision-making. Highlights: GenAI is widely used by HE applicants despite little study of its biases. 216 prompts across 3 chatbots generated 3240 programme suggestions. Grades and priorities drove major shifts in language and recommendations. Gender-coded wording mapped onto research strength and entry standards. GenAI biases may reinforce inequalities in HE admissions decision-making.

Assessment design significantly influences evaluation of student learning. Multiple choice questions (MCQ) and very short answer questions (VSAQ) are commonly used assessment formats, especially in high-stakes settings like medical education. MCQs are favoured for efficiency, coverage, and reliability but may lack depth in assessing critical thinking. VSAQs require students to generate responses, potentially enhancing depth, but posing challenges in consistency and subjective interpretation. Data from parallel MCQ/VSAQ exams over three years was collected. Summary statistics for each exam (marks, time, and discrimination index; DI) and the effect of year and question characteristics were analysed. VSAQs were associated with lower marks (p < 0.001), longer time (p < 0.001), and higher DI (p < 0.001). Question characteristics (e.g. basic science or clinical stems) significantly affected the mark, time, and DI, changing across years, but not interacting with question format. While MCQs resulted in higher marks, VSAQs provided higher discrimination of student performance. Response options in MCQs likely enhance recall, however real-world settings also offer contextual cues. Question characteristics affect student performance independently of format, likely due to differences in cohort career progression. Future research should investigate predictive validity and standard setting of VSAQs in a basic science context.

Stress. What does it mean to you? We probably all refer to it after a long or strenuous day at work, during a particularly heated family argument, or when thinking about our finances. But what does the word really mean, and why is it so important in our everyday lives? Can it increase our risk of developing serious health conditions such as obesity, heart disease, and lung cancer? Is it something that can affect our children as they develop in the womb? Is it even something that can affect our children's children? It may seem surprising, but the answer to all of these questions is yes.

Duplications at 15q11.2-q13.3 overlapping the Prader-Willi/Angelman syndrome (PWS/AS) region have been associated with developmental delay (DD), autism spectrum disorder (ASD) and schizophrenia (SZ). Due to presence of imprinted genes within the region, the parental origin of these duplications may be key to the pathogenicity. Duplications of maternal origin are associated with disease, whereas the pathogenicity of paternal ones is unclear. To clarify the role of maternal and paternal duplications, we conducted the largest and most detailed study to date of parental origin of 15q11.2-q13.3 interstitial duplications in DD, ASD and SZ cohorts. We show, for the first time, that paternal duplications lead to an increased risk of developing DD/ASD/multiple congenital anomalies (MCA), but do not appear to increase risk for SZ. The importance of the epigenetic status of 15q11.2-q13.3 duplications was further underlined by analysis of a number of families, in which the duplication was paternally derived in the mother, who was unaffected, whereas her offspring, who inherited a maternally derived duplication, suffered from psychotic illness. Interestingly, the most consistent clinical characteristics of SZ patients with 15q11.2-q13.3 duplications were learning or developmental problems, found in 76% of carriers. Despite their lower pathogenicity, paternal duplications are less frequent in the general population with a general population prevalence of 0.0033% compared to 0.0069% for maternal duplications. This may be due to lower fecundity of male carriers and differential survival of embryos, something echoed in the findings that both types of duplications are de novo in just over 50% of cases. Isodicentric chromosome 15 (idic15) or interstitial triplications were not observed in SZ patients or in controls. Overall, this study refines the distinct roles of maternal and paternal interstitial duplications at 15q11.2-q13.3, underlining the critical importance of maternally expressed imprinted genes in the contribution of Copy Number Variants (CNVs) at this interval to the incidence of psychotic illness. This work will have tangible benefits for patients with 15q11.2-q13.3 duplications by aiding genetic counseling. The genetic interval 15q11.2-q13.3 on human chromosome 15 contains several so-called “imprinted genes” which are subject to epigenetic marking leading to activity from only one parental copy. This is in contrast to non-imprinted genes, whose activity is independent of their parent-of-origin. Deletions affecting the 15q11.2-q13.3 interval cause Prader-Willi and Angelman syndromes (PWS/AS), depending on whether the deletions are paternally or maternally derived respectively. Duplications at the PWS/AS interval region may also lead to neurodevelopmental disorders, including developmental delay (DD), autism spectrum disorder (ASD) and schizophrenia (SZ). Due to presence of imprinted genes within the region, the parental origin of these duplications may be key to the pathogenicity. We show, for the first time, that paternal duplications lead to an increased risk of developing DD/ASD/multiple congenital anomalies (MCA) but, unlike maternal duplication, do not appear to increase risk for SZ. This study refines the distinct roles of maternal and paternal duplications at 15q11.2-q13.3, underlining the critical importance of maternally active imprinted genes in the contribution to the incidence of psychotic illness. This work will have tangible benefits for patients with 15q11.2-q13.3 duplications by aiding genetic counseling.