Explaining your research in plain language to non-scientists

A quick primer before the walkthrough: Translating science for a non-specialist audience is mostly vocabulary work, and AI is good at vocabulary. The risk is that plain language quietly becomes bolder language. This tutorial builds in the check for that.

In plain English: You paste an abstract you are allowed to share, ask for a version aimed at a named audience, then compare the two for claims that grew.

Who this is for: Scientists asked to explain their work to colleagues in business development, investor relations, communications, or patient engagement — and non-scientists in those roles trying to understand a paper.

What you need: A published or otherwise shareable abstract, and any general AI assistant.

Comfort level: No coding.

Key terms (defined again at the end): preclinical, in vitro, endpoint, mechanism of action.

Step 1 — Name the audience and the one thing they need

What this step does: A summary "for a general audience" is nobody's summary. Naming the reader changes the output more than any other instruction.

Write one line before you start, for example:

• "For a BD colleague who needs to know why this target is interesting to a partner."
• "For a patient-community newsletter, reading age about 14."
• "For a non-technical board member deciding whether to fund the next study."

Step 2 — Draft the plain-language version

Role: You are a science communicator.
Task: Rewrite the abstract below for this audience: [AUDIENCE, ONE LINE].
Rules:
- Use only what the abstract states. Add nothing.
- Keep every hedge: "may", "suggests", "in mice", "in cell culture".
- Do not translate a laboratory or animal result into a statement about patients.
- Define each technical term the first time it appears, in the sentence itself.
- Around 150 words, short sentences, no bullet points.

Abstract:
[PASTE PUBLISHED ABSTRACT]

The hedge rule is doing the heavy lifting. Most overstatement happens when a cautious phrase is dropped for readability.

Step 3 — Build the jargon-swap table

What this step does: Gives you reusable vocabulary, not just one rewritten paragraph.

List every technical term you replaced, as a table:
technical term | plain-English replacement | what is lost in the swap.

The third column matters. "In vitro" becoming "in the lab" is fine; "reduced tumour volume in mice" becoming "shrank tumours" is not, because the species disappeared.

Worked example (illustrative)

Technical sentence: "Compound X inhibited proliferation of three EGFR-mutant NSCLC cell lines in vitro, with an illustrative IC50 in the low nanomolar range."

Plain-language version: "In laboratory dishes — not in people or animals — compound X slowed the growth of three lung-cancer cell lines that carry a specific mutation in a protein called EGFR. It did this at very low concentrations, which is an encouraging early sign."

Step 4 — Run the claim-strength check

What this step does: The final gate before anyone else sees it.

Compare the original abstract and the plain-language version.
List any claim that is stronger, broader, or more certain in the new version.
For each, quote both sentences and suggest a corrected wording.

Then read it yourself against this short list:

• Did a cell or animal result turn into a statement about people?
• Did "associated with" become "causes"?
• Did "may" or "suggests" disappear?
• Did a word like cure, breakthrough, proven, or safe appear from nowhere?

Anything aimed at investors, patients, or the public also needs review by whoever handles that communication in your organization.

If something goes wrong

• It sounds like a press release. Ask for "neutral, factual, no promotional adjectives" and remove every superlative.
• It drops the species or the model. Add: "state the species or system in the first sentence and keep it throughout."
• It is still too technical. Ask for a specific reading level and one concrete analogy — then check the analogy is accurate.
• It adds context that was not in the abstract. Background it supplies from memory is unverified. Delete it, or verify and add it yourself.
• The work is unpublished. Do not paste it into a general assistant — see Safer AI use at work: what to share and what to keep private.

What to try next

• Writing figure captions and lab-meeting slide narratives with AI
• Prompting ChatGPT for high-quality scientific paper summaries

Key terms

• Preclinical — work done before testing in people, in cells or animals.
• In vitro — in cells or components outside a living organism.
• In vivo — in a living organism, usually an animal model at this stage.
• Endpoint — the specific thing a study measures to judge whether something worked.
• Mechanism of action — how a compound is thought to produce its effect.
• Hedge — a qualifier such as "may" or "suggests" that signals uncertainty; keep them.

Plain-language drafts are drafts. Anything going to patients, investors, or the public is reviewed and approved by the people accountable for that communication.