The Noted Scholar

Noted Scholar reviews AI research and writing tools for people who do research for a living — or are trying to survive a degree that demands it. It is written under a pen name by one person who reads and researches daily, much of it with AI assistance, and tests every tool covered here hands-on before writing a word about it.

Who this is — and isn't

Plainly: not a university academic. No affiliation, no lab, no dissertation in progress. What you're reading is an independent tester who lives inside AI reading and research tools every day and documents what they can actually be trusted with. That vantage point has a real advantage — the hallucinated citations, distorted summaries, and confident wrong answers our reviews check for are failures we hit personally, daily — and a real limit: where a tool's value depends on institutional context we can't replicate (submission pipelines, lab collaboration, grant workflows), the review says so rather than pretending. Advice written for PhD students is written for them, not by one.

Why pseudonymous?

Deliberate choice, and worth being upfront about. Publishing under a pen name means this site cannot lean on credentials, an institutional affiliation, or a friendly headshot to borrow trust. Everything has to be earned the harder way: by documenting the testing in enough detail that you can verify it against your own experience with the same tool.

The focus stays on the work, not the person. If a review says Elicit missed three key papers on a literature search, you should be able to run a similar search and see for yourself. That check works whether or not you know the author's name — and it is the only kind of authority this site claims.

How tools are tested

The methodology is fixed, and it applies to every review without exception:

• Real tasks, not demos. Each tool is used on actual work — live literature searches, a real and growing reference library, real long-form drafts including this site's own articles — never on a toy example built to flatter the tool.
• Two weeks minimum before any verdict. Every tool is used for at least two weeks of normal work before a review is drafted. First impressions in this category are consistently wrong, in both directions.
• Failures documented. Every review includes what the tool got wrong: hallucinated citations, missed papers, broken syncs, dark-pattern pricing. If a review ever reads like a brochure, something has gone badly wrong.
• We pay for the tools. Subscriptions are purchased at the normal price. No review copies arranged through vendor marketing teams, no "creator" accounts.

The full version, with the scoring rubric, is in How I Test AI Research Tools.

Editorial independence

No sponsored posts, no paid placements, no "partnerships" with the tools under review. Vendors do not see reviews before publication and cannot pay to change a verdict, a score, or a comparison table. Several of the tools we recommend most often — Zotero, Obsidian, NotebookLM, Semantic Scholar — pay us nothing and never will.

How the site makes money

Affiliate links. When a review links to a paid tool, some of those links are affiliate links: if you subscribe through one, the site earns a commission at no extra cost to you. Three things keep that honest:

• Every article containing affiliate links carries a disclosure at the top, before any link appears.
• Commissions never touch verdicts — we recommend free tools over paid ones wherever the free tool wins on merit.
• The complete policy lives at the affiliate disclosure page.

Corrections

These tools change monthly, and reviews go stale. Every money page is re-verified quarterly, and material changes get an "Updated" date. If you find something wrong — a feature that shipped, a price that moved, a claim that no longer holds — write in and it will be fixed with credit.