What we cover
Teaching, research practice, and the risks of AI tools: invented sources, unequal access, and detectors that flag non-native writers.
Free · No account needed
Research, made clearer.
Plain-language articles and free tools for researchers using generative AI.
Latest articles
Publishing, metrics, and using AI in research.
Writing for publication
Writing for international journals: what “improve the English” usually means
Editorial requests to “improve the English” are among the most common forms of feedback given to authors writing in an additional language, yet the underlying problem frequently concerns what the text is doing rather than how its sentences are formed.
AI in research practiceGenerative AI for novice researchers: a task-by-task guide
Generative AI tools tend to be least reliable in precisely the areas where a user is least able to evaluate their output. This guide is organised around the implications of that asymmetry for researchers early in their careers.
Metrics and publishingWhat SJR actually measures, and why a quartile may mislead
A quartile is not a measure of quality. It indicates a journal’s rank within a field, based on a weighted citation average, and the distinction matters most in the situations where quartiles are most heavily relied upon.
Metrics and publishingSSCI, SCIE, AHCI, ESCI: what your promotion committee actually counts
Three of these indexes are parallel and differ by discipline. The fourth applies different inclusion criteria. Much of the confusion surrounding them, and many of its consequences for careers, arises from conflating these two facts.
AI in research practiceDeclaring AI use in a methods section: wording that withstands peer review
Disclosure statements often fall short in one of two directions: they are too vague to be verified, or so detailed that they imply the model contributed more than it did. Both problems can generally be avoided.
Teaching and learningDesigning assessment when AI detection is unreliable
Current evidence offers little support for relying on detection tools. A more productive question may be what an assessment task actually provides evidence of, and how it can be designed accordingly.
Use the technology. Keep the judgement.
How it's made
Every article is dated, references are checked, and corrections are noted openly. See the editorial policy.
Independent
Free, with no account needed, and not funded by any publisher, indexing service, or vendor. About the project.