{"id":"res-par-group-chat-closure-check","audience":"parents","category":"fact-checking","technique":"verification","title":"Check a group chat message about a school closure against the email","free":false,"task":"A parent posted about tomorrow's school closure in the class group chat, and you want to check it against the district's email before replying. You want exactly 4 numbered lines, one per claim in the post, labeled Confirmed, Wrong, or Not in email, with the email's version after any Wrong label. Use only the district email, plain text, nothing else.","materialLabel":"The district email and the group chat post (attached to your prompt automatically)","material":"EMAIL FROM BAYSIDE SCHOOL DISTRICT (Nov 5, 4:10 p.m.)\nSubject: Lincoln Elementary closed Thursday, Nov 6\nDue to a water main break near the building, Lincoln Elementary will be closed on Thursday, November 6. All other district schools are open. This will be an asynchronous learning day: teachers will post assignments on Seesaw by 9 a.m. Grab-and-go lunches will be available at Bayside Middle School from 11 a.m. to 1 p.m. We will send an update by 6 p.m. Thursday about Friday.\n\n-----\nGROUP CHAT POST (Kendra, 4:45 p.m.)\n\"Heads up!! 1) ALL Bayside schools are closed tomorrow. 2) Assignments will be on Seesaw by 9. 3) Free lunches at the middle school 11-1. 4) School is closed Friday too, they already said.\"","rules":[{"type":"numberedItems","count":4},{"type":"plainText"}],"mistakes":[{"title":"Drafting a reply","why":"The AI writes a polite correction for the group chat instead of checking each claim.","fix":"Say: \"Label each claim only.\"","checks":["r2"]},{"title":"No fixed labels","why":"The AI writes a long comparison, which is hard to scan quickly.","fix":"Give them: \"Confirmed, Wrong, or Not in email.\"","checks":["r0","r1"]},{"title":"Not naming the source","why":"The AI may treat Kendra's post as just as reliable as the district email.","fix":"Add: \"The district email is the only source of truth.\"","checks":["r0","r2"]},{"title":"Not saying how many","why":"The AI decides the length itself and usually returns more lines, one per claim than you can use.","fix":"Be exact: \"exactly 4 numbered lines, one per claim.\"","checks":["numberedItems"]},{"title":"Getting markdown back","why":"AI defaults to bold labels and headings, which show up as stray symbols in a group chat.","fix":"Ask for \"plain text, no bold, no headings.\"","checks":["plainText"]}],"promptHints":[{"label":"Gives exact labels","patterns":["confirmed","not in (the )?email","label"]},{"label":"Names the source","patterns":["only.{0,30}email","email.{0,20}(only|source|truth)","district email"]},{"label":"Asks for the email's version","patterns":["email('s)? (version|says)","what the email says","give the"]},{"label":"Says exactly how many","patterns":["\\b(4|four)\\b.{0,25}(line|claim)","exactly \\d+"]},{"label":"Asks for plain text","patterns":["plain","no bold","no markdown","no formatting","no headings"]}],"hint":"Group chats spread half-right info fast. Make the official email the only judge and give the AI exact labels for each claim.","rubricLabels":["Labels claim 1 Wrong (only Lincoln Elementary is closed; other schools are open)","Labels claims 2 and 3 Confirmed","Labels claim 4 Wrong or Not in email (the email says an update about Friday comes Thursday by 6 p.m.), and outputs only the 4 labeled lines with no intro and no drafted reply to the chat"]}