Students are asking AI which major to choose: fix the bias first, then talk accuracy
The front door to admissions and academic advice is becoming a chat window, and AI answers can carry stereotypes — a risk and an opportunity for institutions.
Key numbers
- Context: choosing a major is hard for incoming students, and traditional advising suffers from long waits, intimidating environments and limited personalisation — which is where chatbots step in.
- Risk: AI systems can generate responses biased by race, gender, socio-economic status and disability, turning prospective students away and undermining reliability.
- The study built a programme-specific advising chatbot for an engineering school and evaluated it with diverse student queries: mean scores of 9.76 for accuracy, 9.56 for relevance and 9.60 for personalisation (out of 10), with no stereotypical bias found in the sampled data.
- The authors note the limits: a small sample and a short timeframe, so real student query variety is not fully covered.
Method and sample
arXiv paper “Bias-Aware AI Chatbot for Engineering Advising” (2025-09-30): a programme-specific chatbot built for an engineering school, tested with diverse student queries and scored on accuracy, relevance, personalisation and presence of bias, with prompt-engineering and mitigation strategies applied.
What it means for brands
For education providers, discovery is moving from websites and directories to students asking AI directly. AI does not only omit you — it can describe you through stereotypes, so whether your official programme facts, entry requirements and outcomes are machine-readable decides which version students see.
How SourceSight handles it
For education projects SourceSight builds prompts from the questions students really ask (which to choose, is it right for me, pricing, credentials) and reports mentions and position per engine; the content factory structures programme details, curricula and requirements.
Limits
The work is a single-school case study with a small sample and a short window; the scores come from the authors' own evaluation rubric and do not generalise to every education setting.
References
Run it on your own project
What the same prompts return across 10 engines is only knowable by running them.
Related research
Cross-category recommendations: +36% add-to-cart on item pages, +15% on cart pages
Recommending general merchandise to grocery shoppers — a milk frother with the milk — has long been hard; LLM retrieval plus real-time cart context moved production A/B metrics.
Industries · 2026-10-07Home services: 22% of homeowners already use AI to find a provider, and 80% of businesses don't know how to be found
The local services funnel is moving upstream: before a homeowner opens a website or calls, AI has already synthesised reviews, reputation and service descriptions into one answer.