Blog

Notes on making AI text read like a person.

No detector-bypass promises — just what actually helps when you’re turning an AI draft into writing you’d put your name on.

14 min read

Why AI Detector Bias Against ESL Writers Cannot Be Fixed

A March 2026 arXiv paper proves the bias against multilingual writers is a mathematical property of text-only detection, not an engineering flaw. Better detectors will not solve it. Here is the proof, the evidence, and what it means for students and institutions.

Read
11 min read

Universities That Disabled AI Detection — Verified Tracker (2026)

Vanderbilt, Waterloo, Yale and Curtin have switched off Turnitin's AI detector, each confirmed on the institution's own page. Widely repeated "50+ universities" counts are not verifiable — here is the difference, with sources.

Read
14 min read

What an AI Detection Score Actually Means — A Guide for Educators

Vanderbilt, Waterloo, Yale and Curtin all switched off Turnitin's AI detector, and the MLA-CCCC task force advises against treating detector output as proof. Here is what a score can support, what it cannot, and how to handle a flag fairly.

Read
15 min read

Falsely Accused of Using AI? How to Prove You Wrote It Yourself

A New York court threw out a 100% Turnitin AI score in February 2026. Peer-reviewed research puts some detectors' false-positive rates on ESL writing above 60%. Here's what actually clears your name — and what wastes your time.

Read
5 min read

How Many Pages Is 1,000 Words?

1,000 words is about 2 pages double-spaced or 4 pages single-spaced in 12pt Times New Roman. Here's the full conversion table, why the answer shifts with font and spacing, and what to check before you trust it.

Read
4 min read

How Many Words Is a 5-Minute Speech?

A 5-minute speech is about 650 words at a standard 130 wpm delivery pace. Here's the full table by duration, how pace changes the number, and why you should write short and time it aloud.

Read
5 min read

Do AI Detectors Actually Work? What the Research Says

AI detectors are treated as evidence in classrooms and newsrooms, but the peer-reviewed data shows accuracy that swings wildly by model, text length, and who wrote it. Here's what the studies actually found.

Read
6 min read

Why AI Detectors Flag Non-Native English Writers

A Stanford study found AI detectors falsely flagged 61.3% of human-written TOEFL essays, while performing near-perfectly on native speakers. Here's the mechanism behind that gap, and what to do if it happens to you.

Read
5 min read

What Is Burstiness in AI Writing?

Burstiness measures how much your sentence rhythm varies. Human writing typically scores 0.6–1.2; model output clusters around 0.2–0.4. Here's what the metric captures, why detectors use it, and why it isn't the whole story.

Read
4 min read

Humanizer vs. Paraphraser: What's Actually Different

Paraphrasers swap words and reorder clauses. Humanizers target sentence architecture and rhythm. The distinction matters because a lot of tools marketed as one are quietly the other.

Read
The Best AI Humanizer Tools in 2026, Tested Honestly
6 min read

The Best AI Humanizer Tools in 2026, Tested Honestly

We ran the same AI-generated paragraph through six humanizer tools and three detectors, and compared price, output quality, and what each tool actually changes. No "100% undetectable" claims — just the results.

Read
6 min read

Is Using an AI Humanizer Cheating? An Honest Answer

Sometimes yes. We sell one of these tools and we're still going to say that, because the answer depends on your institution's policy and on what you're actually asking the tool to do.

Read