Responsible AI 

At Scrolly Science, AI is a tool. It can support our writers, designers, animators and production teams, but it does not replace scientific scrutiny, editorial responsibility or human expertise.

Our approach is simple: use AI where it genuinely improves the work, never where it compromises trust.

Human expertise comes first

Every Scrolly Science project is led by people.

Our writers, editors, designers, animators and production teams remain responsible for the work we create. AI may support parts of the process, but responsibility for the final output always stays with us.

Scientific communication requires judgement, context and an understanding of what the research actually means. We do not delegate that responsibility to a machine.

Scientific accuracy is never delegated to AI

Generative AI can produce convincing information that is incomplete, misleading or simply wrong.

For that reason, AI-generated content is never treated as a scientific source.

Scientific claims are checked against the research, source material and supporting evidence provided for the project. Where appropriate, specialist editorial or scientific review forms part of our production process.

The final responsibility for accuracy remains human.

We are transparent about how AI is used

We do not believe clients should have to guess whether AI has been involved in their work.

Where generative AI makes a material contribution to a project, whether in writing, imagery, animation, voice, ideation or another part of production, we aim to be clear about its role.

Different projects require different approaches.

Some are created entirely through traditional production.

Some combine traditional creative expertise with carefully selected AI tools.

Others may make extensive use of generative technology where the client, subject matter and intended use make that appropriate.

Whatever the production method, our standards for accuracy and accountability do not change.

Your research is treated with care

Researchers frequently trust us with unpublished manuscripts, confidential findings, commercially sensitive information and work that may not yet have entered the public domain.

We take that responsibility seriously.

Client information should not be casually entered into public AI platforms or used in ways that could compromise confidentiality, intellectual property or research integrity.

Where AI tools form part of a workflow, their use must be appropriate to the sensitivity of the material involved.

We do not manufacture scientific evidence

AI-generated imagery can be extraordinarily powerful.

It can also blur the distinction between illustration and evidence.

We will not knowingly present synthetic imagery as if it were genuine experimental data, microscopy, medical imaging, patient information or a real scientific observation.

Where an image is illustrative, it should function as an illustration.

Scientific communication depends on audiences understanding the difference.

Human review remains essential

AI output is not automatically accepted simply because it looks convincing.

Material produced or assisted by AI remains subject to human review.

Depending on the project, this may include reviewing scientific claims, scripts and written content, terminology, references and citations, visual interpretation, imagery, animation, narration, context and meaning, and final presentation.

AI can accelerate parts of the creative process. It does not remove the need for quality control.

Creativity still belongs to people

We work with writers, editors, illustrators, designers, animators, filmmakers and technical specialists because great communication depends upon human creativity.

AI can extend what those people are able to do.

It can help explore ideas, accelerate repetitive tasks, prototype approaches and open up new creative possibilities.

What it should not do is disguise automated output as specialist human work.

Our creative teams remain responsible for the decisions that shape the final result.

Different projects deserve different approaches

We do not believe every scientific communication project should use AI.

We also do not believe AI should be avoided simply because it is new.

The right approach depends upon the research, audience, budget, objectives, confidentiality requirements and intended use of the finished work.

That may mean a completely traditional production process.

It may mean a hybrid workflow where AI supports an experienced creative team.

Or it may mean making substantial use of generative technology under close human direction.

The technology should serve the project, not dictate it.

Trust comes before speed

AI can make some things dramatically faster.

Speed, however, is not the most important measure of good scientific communication.

Accuracy matters.
Context matters.
Clarity matters.
Integrity matters.
And trust matters.

We will continue exploring new technology enthusiastically, but not uncritically.

Our principle is straightforward:

We embrace AI. We do not outsource judgement to it.

Whatever tools are used to create the work, scientific responsibility remains human.