AI Source Finder
Build a source trail for a claim or research question. Mother can help identify the evidence you need, suggest search terms and compare material you provide. It is not a guarantee of live browsing or a verified bibliography: confirm each source directly before citing it.
Start this task in Mother →No account required. The example opens in your composer; edit it before sending. Use uncensored AI for useful work with fewer content restrictions, not a guarantee of an answer.
What to provide
State the question, discipline, date range and what counts as evidence for your purpose. Distinguish primary studies from news coverage and opinion. Supply abstracts or excerpts when you have them and ask Mother to separate what each source establishes from what it merely discusses. For current facts, search the original publisher or database rather than relying on model memory.
Example starting request
I am researching whether later school start times improve teenage sleep. Help me build a source-finding plan for a school policy brief. Prioritize systematic reviews, controlled studies and official guidance. Suggest database search terms, inclusion criteria and a table for study design, sample, measured outcome and limitations. Separate sources I provide from unverified suggestions. Do not invent titles, authors, DOIs or links; say when a citation needs checking.
Worked example
Illustrative example, not a generated result or a user transcript.
For a school-start-time brief, begin with the question: does moving the start later change sleep duration in secondary-school students? Search for “school start time adolescent sleep systematic review” and record the database, search date and filters. In an evidence table, leave the citation and effect-size cells blank until you have opened a real paper. Separate measured sleep outcomes from opinions about attendance or convenience.
This is a search method, not a claim that any particular study proves the policy works. A before-and-after study without a comparison group may confuse the timetable change with other changes. Prefer directly relevant outcomes and record how they were measured. If an abstract omits a limitation, mark it as unknown rather than filling it from an AI summary.
Check the answer
Open each original source and verify its title, author, publication date and identifier. Read the methods and limitations, not just the abstract. Check for corrections or retractions and record the exact passage supporting your claim. Agreement between several summaries may trace back to one weak study. A plausible citation is not proof it exists; if you cannot find a proposed reference, exclude it until verified. When research tools are enabled, Mother can use public web search and page reading; availability and access to a particular page are not guaranteed. Ask which pages were actually read and which findings came only from search snippets. A blocked or truncated page is not a checked paper. If a search finds only coverage, trace its named study to the original publisher, or mark the claim unsupported. Evidence-row template: citation [verified identifier or not found]; design [from methods]; population [from source]; outcome [measured variable]; limitation [stated or unknown]. These placeholders are not a citation or an observed study result.