Required from Week 2 · One notebook, all fifteen-plus sources, all semester
NotebookLM holds every article you collect — and answers questions using only those articles.
NotebookLM (notebooklm.google.com) is a free research tool from Google with one property that makes it different from every chatbot you've used: it is grounded. You upload your sources — the actual PDFs of the articles in your literature review — and it answers questions using only those documents, citing the exact passage every claim came from. Ask ChatGPT about your topic and it answers from the entire internet, sometimes inventing citations that don't exist. Ask your notebook and it answers from your fifteen articles, with a footnote you can click to see the original sentence.
That property matters here for three reasons. First, your sources live in one place. By Week 7 you will have fifteen or more PDFs; by finals week, likely twenty. Scattered across downloads folders and email attachments, they rot. In one notebook, they are searchable as a collection — which is exactly how a literature review has to treat them. Second, it lets you interrogate your literature as a whole. The skill of Week 4 — synthesis, seeing across studies instead of summarizing them one at a time — is precisely what querying a full notebook rehearses. Third, every answer is checkable. The citation takes you back to the source, and going back to the source is the habit this course exists to build.
A notebook does not read for you, and it does not write for you — the AI policy below draws that line in bold. It is a filing cabinet that answers questions and shows its receipts.
1 · Sign in with your SJNY account. Go to notebooklm.google.com and sign in with your SJNY Google account — the University's license covers NotebookLM, and the course requires the SJNY account because your notebook must be shared within the sjny.edu domain (step 5). The source limit is far more than this project needs.
2 · Create one notebook. Click "Create new" and name it PSY 460 Capstone — [Your Topic]. One notebook for the whole semester. Do not make a new notebook per week or per assignment; the whole point is that your literature accumulates in one place.
3 · Upload your PDFs. As you collect articles (starting with your ten candidates in Week 2), add each full-text PDF as a source. Upload the full text, not the abstract — abstracts are advertisements, and a notebook full of advertisements answers like one.
4 · Rename each source to its citation. NotebookLM names sources by filename, and fulltext(3).pdf tells you nothing. Rename each source to its short APA citation — Neff (2023), Baumeister & Leary (1995) — so that when your notebook cites a source, you instantly know which paper it means. This small habit also drills the author–year format your reference list depends on.
5 · Share it with me. Click Share and add mwmagee@sjny.edu as Viewer — required, by the end of Week 2. This mirrors your seminar Docs, just in reverse: the notebook lives on your account, and I have a window into it. Viewer access means I can see your source list and ask your notebook questions while preparing for our meetings — but I cannot change anything.
6 · Keep it current. Every article that enters your synthesis matrix enters your notebook the same day. "How many sources are in your notebook?" is no longer a question I have to ask — I can see for myself, and a notebook that grows steadily across the semester tells the same story your journal does.
Week 2 — collection. Upload your ten candidate articles. Then ask: "What research questions do these sources address? What methods do they use?" Ten minutes with the answers tells you whether your candidates form a literature or a grab bag — before your topic locks.
Week 3 — dissection (in that order). Complete your article-dissection worksheets yourself first. The worksheet is the skill; outsourcing it teaches the tool and not you. Then ask the notebook about the same article — "What were the IVs, DVs, and key limitations of Smith (2021)?" — and compare. Where the notebook surfaced something you missed, go back to the PDF and find it. That comparison is a comprehension check you can run on every article all semester.
Week 4 — synthesis. This is where the notebook earns its keep. Ask cross-cutting questions: "Which sources measure self-esteem, and how do their measures differ?" "Where do these studies disagree?" "What populations has this literature ignored?" The answers are raw material for your synthesis matrix — leads to verify against the PDFs, not rows to copy. The matrix you submit must be built, and worded, by you.
Weeks 5–6 — the gap and the outline. Pressure-test your gap statement: "Do any of my sources already test whether X affects Y?" If the notebook finds one that does, better to learn it now than from me. As you outline, ask which sources speak to each planned section — a fast way to spot the section resting on a single reference.
Week 8 — measures. Ask: "What instruments do my sources use to measure [construct], and what reliability do they report?" Then pull the actual numbers from the PDFs for your scales table — the notebook points, the source provides.
Weeks 7, 10–12 — while writing and revising. The notebook is your fact-checker: "Did Neff (2023) actually claim this, or am I remembering it wrong?" Checking a claim against the source before you cite it is the single most professorial habit you can steal. What you may not do is have the notebook draft your prose — see the line below.
This course's disclosed-use AI policy applies to NotebookLM exactly as it applies to any other AI tool. The notebook's answers are AI-generated summaries — and submitting AI-generated summaries as your own writing is never permitted, in any draft, worksheet, journal entry, or matrix. Do not paste notebook output into your seminar Docs. Do not lightly reword notebook output and type it in; paraphrasing a machine is not writing.
What NotebookLM is sanctioned for: storing and organizing your sources; locating which source said what; checking your comprehension after you've done the reading; asking cross-cutting questions whose answers you verify in the PDFs; and fact-checking your own claims against your own sources before you submit. Used this way, it is precisely the kind of "defined support task" the policy permits.
Log it. Substantive notebook queries — anything that shaped your thinking, your matrix, or your draft — get an AI log entry like any other AI use: date, tool ("NotebookLM"), what you asked, what you did with the answer. "Asked my notebook which sources measured anxiety; verified in the PDFs; added two rows to my scales table" is a perfect entry, and an honest log entry is never penalized. Routine housekeeping (uploading a PDF, renaming a source) needs no entry.
One more reason this arrangement protects you: your writing happens in seminar Docs whose revision history I can see, and your notebook use is disclosed in your log. A student who reads deeply, queries their notebook, verifies against sources, and drafts in their own words in their own Doc has nothing to fear from any integrity question — the whole trail shows the work is theirs.
It cannot find literature. A notebook only knows what you feed it. Searching is Week 2's skill, and it happens in PsycINFO and Google Scholar — not here.
It cannot replace reading. Our weekly meetings are conversations about your articles, and thirty minutes of questions reveals within about ninety seconds whether your understanding comes from the papers or from summaries of the papers. The notebook checks reading you have done; it cannot substitute for reading you haven't.
It can still be wrong. Grounding makes fabrication rare, not impossible — the notebook can misread a table, over-generalize a finding, or miss the qualifier that changes a claim's meaning. That is why every citation is clickable: click it. Never move a notebook-sourced claim toward your paper without seeing the original passage with your own eyes.