Top Headlines

Feeds

Swedish University Drops Fagessay Requirement, Proposes Workshops to Preserve AI‑Resistant Assessment

Updated (2 articles)

Remote assessments erode verification of student authorship The shift to remote exams makes it impossible to confirm that students actually wrote their submissions, raising acute concerns about undetectable AI‑generated work. Both oral and written formats present problems; oral exams increase stress and may hinder performance, while written exams remain unpredictable due to variable tasks and timing. Institutions therefore seek alternatives that can assure authenticity. [1]

Fagessay structured to thwart generic AI output The fagessay requires students to weave personal practice stories with theoretical reflection, producing a highly individualized narrative. Its lack of a fixed template and demand for case‑specific insight prevent generic chatbot text from satisfying assessment criteria. Proponents argue this format preserves academic integrity in an AI‑rich environment. [1]

HVL eliminates fagessay requirement citing privacy law Høgskulen på Vestlandet (HVL) recently removed the fagessay component from several programmes, stating that personal narrative elements clash with emerging privacy regulations. The policy change reflects growing legal pressures on institutions to protect student data while maintaining assessment standards. [1]

Peer‑guided workshops planned to rebuild authentic writing skills In response to the policy shift, HVL proposes essay workshops where students collaboratively develop and critique fagessays, aiming to restore confidence in producing KI‑free work. These sessions are intended to reinforce the pedagogical value of reflective writing and ensure continued resistance to AI‑assisted cheating. [1]

Sources

Timeline

2020s – The widespread shift to remote assessments “makes it impossible to confirm that students authored their papers,” raising acute academic‑integrity concerns about undetectable AI‑generated work[2].

Fall 2025 – Virginia Tech launches an AI‑powered essay reader that “speeds admission decisions by about a month” and scans 250,000 essays in under an hour, saving an estimated 8,000 staff hours while flagging any AI‑human score gap over two points for secondary review[1].

Fall 2025 – Caltech rolls out an authenticity‑check chatbot that “interviews applicants about their research projects to gauge authenticity and passion,” with faculty reviewing the video responses to verify genuine intellectual claims[1].

Fall 2025 – UNC Chapel Hill implements AI‑driven grammar and style evaluation of applicant essays, prompting backlash as applicants and parents claim the tool “provides data points but does not replace comprehensive human review”[1].

Fall 2025 – Georgia Tech deploys AI to automatically read transfer student transcripts, eliminating manual data entry and “reducing the time needed to determine credit‑transfer eligibility” while also flagging low‑income Pell‑Grant candidates[1].

Fall 2025 – Stony Brook University pilots an AI system that reviews transcripts and summarizes essays and recommendation letters, highlighting contextual factors such as illness or family responsibilities to inform admissions decisions[1].

Fall 2025 – The National Association for College Admission Counseling adds a dedicated AI section to its ethics guide, urging institutions to use AI “in ways that align with transparency, integrity, fairness, and respect for student dignity”[1].

Feb 25, 2026 – Høgskulen på Vestlandet (HVL) drops the fagessay requirement from several programmes, citing “conflicts between personal narratives and privacy regulations” as the reason for the change[2].

2026 (planned) – HVL proposes peer‑guided essay workshops aimed at “rebuilding confidence” and helping students produce authentic, AI‑free reflective essays, reinforcing the learning value of writing[2].

2026 (ongoing) – Advocates promote the fagessay model as an AI‑resistant assessment, noting that its “lack of a fixed template and requirement for personal reflection on specific cases makes generic AI‑generated text struggle to meet the criteria”[2].

All related articles (2 articles)

External resources (5 links)