Generative AI · Education · Research
An independent, scholar-led platform for educators and researchers working out where generative AI belongs in their practice. Plain-language writing on publishing, metrics, and research method, alongside free tools for the checks that come before a submission—enabled by AI, but directed by human expertise, ethics, and care.
01 · Why ELT Pub
Generative AI changed what a machine can produce. The harder question is what it has changed about how we learn, how we teach, and what we accept as knowing something.
The gains are real and worth naming. A student can ask for an explanation at midnight, then ask for a different one. A teacher can adapt a reading for three levels in the time one used to take. A researcher can find the literature they did not know to look for. Used with care, that is a genuine widening of access and attention.
But fluency is not truth, and speed is not wisdom. The same systems invent citations with total confidence, carry the biases of what they were trained on, and are least reliable precisely where a newcomer is least equipped to notice. ELT Pub takes both halves seriously: what these tools do well, and where they fail.
02 · What the site covers
Three subject areas, spanning education, applied linguistics, and the social sciences.
Where these tools help a learner, and where they quietly do the learning for them. Feedback, explanation, materials design, and assessment, examined for what a student gains and what they stop practising.
Pedagogy · Evidence · PracticeLiterature searching, synthesis, analysis, and drafting are the tasks AI is sold hardest for. We set out which of them survive contact with peer review, and what each one still demands you verify yourself.
Inquiry · Methods · CommunicationInvented sources, unequal access, unclear authorship, detection tools that misfire on non-native writers. The costs are easier to overlook than the benefits, which is exactly why they get their own coverage here.
Ethics · Risks · Human agency03 · The writing
Explanatory and critical pieces on research metrics, publishing integrity, and where generative AI genuinely helps a working researcher.
The most common feedback given to non-Anglophone authors is also the least useful, because the underlying problem is usually not the sentences. It is what the sentences are being asked to do.
AI in research practiceThese tools are least safe exactly where you know least, which is an awkward property for something marketed to beginners. That asymmetry is what this guide is organised around.
Metrics and publishingA quartile is not a quality score. It is a rank within a field, derived from a weighted citation average, and the difference matters most in exactly the situations where people rely on it.
Metrics and publishingThree of these indexes are peers that differ by discipline. The fourth sits outside the Core Collection. Most confusion — and most career damage — comes from conflating those two facts.
AI in research practiceMost disclosure statements fail in one of two directions: too vague to be checkable, or so detailed they imply the model did work it did not do. Both are avoidable.
Teaching and learningThe detection question is settled badly enough that it is worth stopping asking it. The useful question is what an assessment is actually evidence of.
04 · The toolkit
Four free utilities, none of which ask you to create an account. Two of them run entirely inside your browser and never send anything back to us.
Search a journal by name or ISSN for an integrity-risk profile drawn from publication growth, authorship concentration, and journal-title warning lists — with the caveats stated alongside the result.
Open tool ↗ DiscoveryA browser extension that puts Scopus quartiles, SCImago indicators, and Web of Science indexing badges directly onto your Google Scholar results, so you can judge a venue without leaving the page.
Open tool ↗ FocusAn analogue countdown whose red sector grows as the session is spent, paired with an independent stopwatch for timing tasks, observations, or presentation runs.
Open tool ↗ ProductivityDated tasks with due dates, priorities, and progress, held in your own browser's storage. Nothing is transmitted, nothing is stored by us, and no sign-in is possible.
Open tool ↗05 · Using AI in practice
The same three-step habit covers most honest uses of these tools, in a classroom or in a manuscript. It is not a restriction on what you can do with them. It is what keeps the time they save from being spent later, correcting something you did not check.
Let the model widen the search, draft the scaffold, or surface a pattern you had not considered.
Open every source it cites. Test the claim against something it did not write. Ask what it left out.
Interpretation, ethics, and anything you will have to defend stay with the person whose name is on the work.
06 · Editorial standards
The previous section is what we ask of a reader. This is what we ask of ourselves — the rules that govern what gets published here, set out in full in the editorial policy.
Every article carries a named human author, a publication date, and a last-updated date where it has been revised. There are no house-name, pseudonymous, or unattributed pieces on this site.
Every reference is checked to exist and to say what it is claimed to say. Language models fabricate plausible citations, and a fabricated citation in an article about research integrity would refute itself.
A factual error is fixed with a note stating what was wrong and what it now says. The mistake is described rather than erased, and an unsupportable article is marked retracted and left in place.
Generative tools draft scaffolds, tighten prose, and review code for this site. They do not generate factual claims, supply citations, or make editorial judgements — and the editorial policy lists both halves explicitly.
07 · Intended impact
A deliberately small catalogue, built to be useful rather than large.
“Make scholarship easier to discover, harder to misuse, and more valuable to society.”
The measure is not how much gets published here. It is whether a reader leaves able to ask a better question, judge a source more accurately, or make a decision they can defend to a committee.
Long-form and explanatory, on metrics, publishing integrity, and research practice. Each one lists its reading time before you commit to it.
Two of them never transmit anything at all. None asks for an account, and none is funded by the organisations whose data it reports.
Built for curious, careful minds
6 articles, 4 working tools, and one researcher's attempt to be useful about a technology that is changing academic work faster than the advice about it can keep up.
08 · Founder & stewardship
ELT Pub is designed, developed, written, and solely sponsored by Dr. Ngo Cong-Lem. It takes no payment or consideration from any publisher, indexing service, university, or software vendor, and holds no commercial relationship with the organisations whose data its tools report. The work draws on his research across AI in education, applied linguistics, research methodology, and evidence synthesis.
Independent · Founder-led · Not-for-profit in spirit