Field notes from the research desk
Noted Scholar field-tests AI research assistants, reference managers, and academic writing tools on real papers, real literature reviews, and real deadlines — then reports what worked, what broke, and what is not worth your grant money.
The promise is simple: every tool reviewed here has been used on actual research for at least two weeks, and every failure we hit is documented alongside the strengths. No demo-driven reviews, no vendor copy, no hype.
What we cover
Pillar 01
Elicit, Consensus, Scite, SciSpace, Perplexity — every one of them claims to search the literature better than you can. Some of them occasionally do.
We run each tool against real research questions with known answers, count the hallucinated citations, and tell you which ones earn a place in an actual workflow.
Pillar 02
Your reference manager will outlive every AI tool on this site. Zotero, Mendeley, EndNote, Paperpile — the choice you make here follows you through your whole career.
Setup guides that go deeper than the library guides, migration walkthroughs that do not lose your annotations, and a clear answer on which one to pick.
Pillar 03
Writing tools are where the money and the ethics collide. Paperpal, Jenni, QuillBot, Grammarly — each is useful in a narrow band and a liability outside it.
We review the tools on real long-form drafts and cover the part vendors skip: what your university’s AI policy actually permits, and how to disclose what you used.
Pillar 04
Tools are only half the problem. The other half is the system: how a paper goes from PDF to annotation to a note you can actually find when you write.
Full setup guides for the Zotero + Obsidian + Readwise pipeline, and honest assessments of which productivity systems survive contact with a real PhD.
Pillar 05
The literature review is where AI tools promise the most and where uncritical use costs you the most. A summarizer that skips the method section is a trap.
Step-by-step review workflows, summarizer head-to-heads run on identical papers, and guidance on what is legitimate in a systematic review in 2026.
Latest field notes
Flagged for AI you didn't use? Don't confess and don't rewrite. Gather version history, drafts, and notes, respond in writing, and know the appeal path.
Using QuillBot isn't plagiarism by itself: plagiarism is about uncredited ideas, not tools. The real lines are citation, permitted assistance, laundering.
Is AI allowed in academic writing? What Elsevier, Springer Nature, IEEE, NIH, and 2026 university policies actually permit — and how to disclose it.
How to annotate research papers so the marks feed your notes: color conventions, marginal comments, and why Zotero's reader beats iPad apps in 2026.
Notion vs Obsidian for research notes in 2026: verified pricing, where each wins, the data-longevity case, and the hybrid setup many researchers run.
Who writes this
The Noted Scholar is a pen name — deliberately. The persona stays anonymous so the work has to stand on its own: no borrowed credentials, no institutional halo, just documented testing you can check against your own experience.
What you can hold us to is concrete: