Fabricated Citations After the Chatbot
A 10th-grade English teacher brings her class to the library for a persuasive research paper. The librarian notices that 11 of 28 students have cited at least one source that cannot be found, including a fake article attributed to a real university and a statistic that appears only on an AI-generated summary site. The teacher is furious and wants to ban all AI tools and require handwritten note cards. A quick student survey shows that 40% used a chatbot at some point, mostly to 'find sources fast.' The librarian knows the district's acceptable use policy allows AI for brainstorming but requires verification of sources. The unit ends in three weeks, and the teacher has already spent two weeks on thesis statements. The librarian has two 45-minute blocks with the class before the first draft is due. Students are engaged but defensive; several say they did not know chatbots could invent citations. The teacher is open to a short intervention but does not want to 'reward cheating' with extra time.
1. How should the librarian diagnose the problem before designing instruction? What evidence would distinguish a citation-verification deficit from a motivation or time-management issue?
Read the worked answer
Before teaching, the librarian should separate three possible problems: students cannot verify sources, students do not understand what a chatbot does, or students are rushing and taking the easiest path. She can analyze the 11 bad citations by type-fabricated author, real organization but fake article, misattributed statistic, dead URL-and compare them with students' notes and search history if available. A five-minute anonymous survey can ask where students found each source, whether they opened the original, and how they checked the author. She should interview four or five students, including some with accurate citations, to compare process. If students can verify when prompted but did not, the issue is habit and time; if they cannot explain provenance, it is a skill gap; if they knew the source was weak but used it anyway, it is motivation or grade pressure. This diagnosis matters because banning AI would not teach verification. The intervention should target the actual gap, and the teacher's concern should be addressed with evidence rather than blame.
2. Design a two-block intervention that addresses AI-generated source errors and builds durable verification habits. Include specific activities and formative checks.
Read the worked answer
Block one: use a short AI-output example with a plausible but fake citation. Model a source trail: click the link, search the author and title independently, check whether the publication exists, and look for the statistic in at least two reliable sources. Teach lateral reading-leave the site and see what others say about the source-and a 'verify before cite' checklist. Students then audit one of their own sources in pairs and mark it green, yellow, or red. Block two: students revise one citation and write a three-sentence provenance note. The librarian models a chatbot prompt audit: ask the tool for sources, then verify each; if it cannot provide a traceable source, do not use it. Formative checks: exit ticket with one verified and one rejected source, plus a pair conference. The reasoning is to make verification visible and repeatable, not to lecture about cheating. The teacher's three-week timeline means the intervention must improve the current draft, so revision is part of the lesson, not an add-on.
3. How should the librarian and teacher assess whether the intervention worked and communicate results without punishing students for past errors?
Read the worked answer
Assessment should focus on improvement, not punishment. The librarian and teacher can compare the first draft citation set with a revised set using a simple rubric: source exists, author is credible, statistic is traceable, and citation format is accurate. A pre/post count of unverifiable sources gives a credible measure. They should also collect a short reflection in which students explain how they checked one source. To avoid punishing past errors, the teacher can allow a revision window and treat the first draft as formative if the assignment policy permits. If the district policy requires consequences for academic dishonesty, the librarian should help distinguish between using AI for brainstorming, which is allowed, and presenting fabricated sources as real, which is a citation error that can be remediated. Communication should be private and specific. The reasoning is to preserve trust, because defensive students are less likely to change habits; public shaming would likely increase concealment rather than verification.