Four Opportunities for Generative AI in Science Communications

Thinking beyond content generation can make generative AI more useful for science writing and communications.

Like many other creative fields, science communications is skeptical of generative AI, and for good reasons. The communication limitations of large language models (LLMs) are well-publicized at this point: concerns around training data, generic outputs, and factual inaccuracies are why a number of science comms organizations and publications, including notable journals like Science, Nature, and the National Association of Science Writers, have released editorial statements against the use of AI-generated content and imagery. 

We get it. Several of us are NASW members, and have an extensive background in science and tech communication. In addition to earning PhDs in science and scientific and technical writing, we have led strategic content campaigns and taught innovation and data storytelling training for innovative brands in science, tech, and medicine for over a decade. We have seen through our work and research how navigating new media and technology can put pressure on communicating science effectively and accurately.

When technologies change how we create and communicate, they inherently impact how audiences perceive, understand, and engage with science too.

But we have also seen how science comms innovators have adopted new technologies to forward knowledge and engage publics in a variety of imaginative ways - and like any other company, we are actively experimenting with generative AI tools to make our work better, faster, more creative, and more impactful. 

In this article, we share four opportunities for using generative AI responsibly as science communicators. Read on to learn how to use generative AI to:

• Accelerate research and interdisciplinary inquiry
• Generate not just better writing, but better science
• Make impactful science comms skills more accessible to scientists 
• Help the globe hear from the diverse world of scientific inquiry  

‍

Get smart fast—and thus innovate across disciplines.

From Kate, PhD in English, Histories of Science and Tech 

Science communicators often need to develop knowledge and authority in new areas of study - I love using generative AI as a research assistant for that work. Navigating a slew of PDFs in a field or on a topic in which I need to get smart fast is much quicker with AI generated summaries and chatbots helping collect, scan, and compare secondary source material. Developing a comparative overview of a densely researched field is now more efficient with generative AI. More high-quality databases and platforms (including Narratize) are launching AI-enabled research assistance, so these tools are increasingly accessible and worth a try.    

And here’s an even more exciting opportunity: researchers can now afford to specialize and (reasonably) generalize, spurring opportunities for interdisciplinary inquiry and collaboration. So many of our world’s challenges require interdisciplinary solutions - the research projects pitched to the United Nations World Food Forum challenges are a case in point - and generative AI makes it easier for a specialist to pursue meaningful expertise in related fields. 

Check: Any summary is prone to inaccuracy, and reading AI generated summaries is not a replacement for reading articles for closer detail. Finding the most relevant source material can’t be entirely outsourced to generative AI either - authors should still do due diligence looking through reference pages, checking on number and quality of citations, taking courses when needed, and communicating with colleagues to stay at the cutting edge of their fields. 

‍

‍

Use Writing as a Tool for Research and Discovery

‍From Kendall, Ph.D. in Technical Writing and Communication

GenAI is an epistemic technology that can help scientists and writers alike explore ideas, test concepts, and fine-tune a project. 

When we think of science communication, we often think of the writing workflows characteristic of the end of the innovation cycle: sharing findings that detail discoveries, the documenting and polishing of ideas to share with readers. 

But when we focus on the end result, we’re missing out on the potential for GenAI to facilitate the generative aspects of writing. In my experiences working with scientists, there is a lot of useful intellectual work happening through writing that is often overlooked and underappreciated. This is why many people whose professional roles don’t seem to be writing intensive may not see themselves as writers. Or they see GenAI as a technology that’s useful only if and when findings are ready to be shared with a wider external audience.  

I see the potential of GenAI to reframe science communication as not only about communicating about science, but also communicating with science. This goes beyond creating accurate outputs and extends the potential of generative AI-enabled communication to research and discovery. 

Check: To do this work well, generative AI needs to be usefully and strategically adopted into writing and research workflows. That means seeking opportunities to involve generative AI technologies into the communication practices that are inherent to research activities, product development, and R&D. In the corporate world, there is a lot of opportunity for this work: scientists, engineers, and technologists are already spending 30% of their time writing and communicating. But every organization is different, and humans should lead the way in identifying opportunities and testing the impact of applications.  

‍

Make science communications more accessible to scientists. 

From Alicia, Ph.D. in Chemistry

Scientists and engineers are incredible at what they do–they push the bounds of understanding in their respective corners of their fields. For efficiency and out of necessity, they lean on highly technical terminology to communicate with their peers in their field. While that way of communicating works when experts are speaking among themselves, research requires funding and resources. To do this, scientists must transform complex concepts and data into actionable language. And while many universities, organizations, and research centers are paying for science comms training, there is much more work to be done to scale these skills within the scientific community.

Generative AI allows experts to remain experts in their own fields while still doing this critical work. A cancer researcher doesn’t have to become a technical writer or narrative scientist in order to win funding or green-light next quarter’s research projects. It holds the potential to make effective science communications more accessible to scientists by translating technical jargon into actionable language that everyone understands. Identifying the best way to tell the story–as my grad school advisor told me, chronological order is not typically the most impactful–is crucial to this work. The generative AI we are building into Narratize can help identify the highest impact storytelling strategies (patterns, techniques, frameworks) for a given audience and innovation context. 

Check: Equipping scientists with technologies that prioritize accuracy, impactful science storytelling, and IP protection is critical. If scientists don’t trust the product, they won’t use it. And if they don’t own the copyright over AI co-authored texts, they can’t publish it either. Vendors who acknowledge and understand these roadblocks will be critical to ensuring that generative AI empowers scientific communication skill-building. 

‍

Help science hear from local voices and diverse teams. 

From Katie, PhD in Rhetoric and Technical Communication 

It’s no secret that English is the lingua Franca of science, and it is so because of systems of power that have prioritized English-speaking people and research delivered in English. LLMs are poised to reinforce this imbalance. But as the LLM landscape continues to include more guardrails, languages, and translation abilities, science writers and scientists making smart use of generative AI will be able to communicate in their chosen language - and translate effectively for different audiences - without needing to develop fluency in multiple languages. 

Check: Translation is a very complex literacy task, and without close attention and oversight, translations can easily introduce inaccuracies. Translation work requires at least a two-eyes review policy. Science writing organizations can help by pairing strong translators with research teams to support this work. Because great ideas can come from anywhere, and science communication exists to amplify bold ideas across the globe. 

‍

‍

At Narratize, we envision a world illuminated by brilliant ideas. Learn more about how our generative AI platform is empowering brands with innovative ideas to unleash their full potential.