Quality Assurance Framework

Writing and reviewing content

AI generates course content from a structured curriculum and stated learning outcomes in English and Turkish. Automated checks are used to identify accuracy problems, inconsistencies, banned claims and unsupported qualification or accreditation wording. These checks can miss errors and are not independent expert approval.

Every course is reviewed at least once a year. The review checks accuracy, legal and food safety references, clarity, and the link between lessons and assessment. Each course page shows its "Last reviewed" date once a review has been recorded.

Corrections reported by learners or staff are made between reviews. Earlier text is kept so changes can be traced.

Monitoring exam questions

Final exams draw 30 questions at random from a larger pool, with answer options shuffled for each attempt. Once a question has at least 30 graded answers, we calculate its facility index (the share of learners who answer correctly) and its discrimination index (how well it separates stronger from weaker candidates).

Questions that are too hard (facility under 0.30), too easy (over 0.95) or do not discriminate (under 0.15) are flagged for review. They are rewritten or withdrawn, and every change is logged. A course always keeps at least 30 active questions.

Each exam starts with an integrity declaration. Leaving the exam window is counted and reviewed by staff; it never fails a learner automatically.

Using learner feedback

After earning a certificate, learners are asked five short questions on content quality, clarity, relevance to work, assessment fairness and overall rating, with space for comments.

Results are reviewed for every course as part of the annual review. Low scores or repeated comments lead to changes. A course page shows its average rating only when at least 10 learners have answered, so that the figure is meaningful. We never publish invented ratings or testimonials.

Planned external review

We plan to invite an independent, suitably qualified reviewer to examine a sample of courses and assessments, and to publish a summary of the findings. No external review has taken place yet. When one does, it will be listed on our Recognition page.

AI reports and human decisions

Learners can use Report this answer on completed AI Trainer replies. The reported answer and reason are stored privately for admin review. AI reports, course feedback and exam item analysis help identify problems; they are not a guarantee of accuracy. No AI report statistics are published without verified data.

Murat Bilican, Founder, is accountable for academic decisions. Human review is required for appeals, malpractice cases, certificate revocations and final diploma results. Any automated result can be referred for free human review through Contact, topic Assessment appeal, within 20 working days.