A source link is a starting point

What we found in the news

A September 2023 study in Scientific Reports (opens in a new tab) examined references produced by two versions of ChatGPT for short literature reviews. The researchers found invented references as well as errors in citations to real publications. They described the first problem as:

“fabricated bibliographic citations that do not represent actual scholarly works”

— William H. Walters and Esther Isabelle Wilder

The study tested particular models and prompts; it does not describe every tool or today’s performance. It gives us a useful starting point for reviewing a resource: a reference that looks complete still needs checking.

Why this matters to us

Imagine a coordinator preparing a short resource about joining community activities. An assisted draft says a particular kind of group improves confidence and adds a journal citation. The coordinator finds a real paper with a similar title, but it concerns a different group of people and a different activity.

The citation exists. The proposed claim still needs work.

At genus, we want supporting reading to help someone understand an idea and decide what it means for them. A source should make that conversation clearer. It should also make it possible to see where our interpretation begins.

Open the source before carrying the claim forward

Follow the link and compare the publication’s title, authors, and date with the draft. A link to a search page or a journal’s home page may require further work before anyone can assess the claim.

Then find the passage that is supposed to support the sentence. Does it concern the same activity, audience, and situation? Is the writer describing measured results, participants’ opinions, an early proposal, or a review of other research?

In our imagined resource, the coordinator might discover that participants said they enjoyed attending. That is useful information. It would not, on its own, support a claim that the program caused a lasting change in confidence.

Keep the language close to what was actually established. A modest, accurate observation gives the reader more to work with than an impressive sentence whose support disappears on inspection.

Let the unanswered question stay visible

Sometimes a source is relevant without answering the whole question. A study may leave out people who need a particular communication format or may describe only a short trial.

Write down that limit while it is fresh. It may belong in the article, a reference note, or the team’s review record. Avoid asking an assistant to smooth away the uncertainty just because the paragraph feels unfinished.

A participant could help here by asking a question the draft did not consider: could someone join occasionally, bring a supporter, or watch before taking part? Those questions might lead the team toward practical information instead of another general benefit claim.

The source can inform an invitation while leaving room for what the local community still needs to learn.

Keep a useful trail for the next reviewer

Save the direct source link, the publication date, and the particular point it supports. If the team has changed a claim, record why in the appropriate review notes.

That makes a later update easier. A colleague should not have to reverse-engineer a confident paragraph to discover which sentence came from research and which came from the organization.

Check links again when reusing an older resource. A changed page, a corrected paper, or a new version may alter what the team should say. The review trail helps a colleague notice that change without treating every resource as a fresh investigation.

If support cannot be found, remove or narrow the claim. There may still be a helpful article to write around the question itself.

What this means for the people you serve

Supporting links should give people a fair chance to inspect the reasoning. Explain unfamiliar study language and identify limitations that affect the reader’s choices. People can contribute questions and challenge an interpretation without needing to become research specialists first.

The genus connection

Today, configured genus care apps and care walls can present prepared articles, useful links, and activities. Staff choose and organize those resources for the intended audience.

That provides somewhere to share reviewed information. It does not establish the reliability of every linked claim. We would keep source checking within the content-review work and confirm the available authoring and access arrangements with the organization.

One question to take away

If someone follows the reference behind our most important sentence, will they find what that sentence promises?