33. AI For Science Communication: Prof Mike S. Schäfer on AI's promises, pitfalls, and pink slime

AIConspiracy theoriesTrustCommunicating in different formats/mediumsUser profiles/personas/archetypesStakeholder/audience mappingRepresenting your audience's POV
Written By Chris Pahlow

Subscribe to the Amplifying Research podcast:
Apple Podcasts | Spotify | Overcast | Pocketcasts

Episode show notes

How is generative AI transforming the way research is communicated and understood by different audiences? Prof Mike S. Schäfer gives us a state of the union on the rapidly evolving world of AI and science communication.

Mike is a professor of science communication, the director of CHESS (Centre for Higher Education and Science Studies), and head of the Department of Communication and Media Research at the University of Zurich.

"In science communication, dialogue often, by many, is seen as the best way of doing science communication, but the challenge is having a dialogue with people who are not interested in science to begin with... It's difficult to scale up if you actually want to do face-to-face dialogue. And AI is great at that." -- Mike Schäfer

Whether you're a research leader looking to build your team's communication capacity or an individual researcher seeking to develop your skills, this episode offers practical advice on making training work for the long term.

Our conversation covers:

  • How the general public is increasingly using tools like ChatGPT to get answers about science

  • The characteristics of early AI adopters and concerns about digital divides in AI literacy

  • How AI hallucinations and "pink slime" could impact scientific knowledge

  • The ways that ChatGPT and other models conceptualise and communicate science

  • Practical applications for researchers to experiment with AI in their communication efforts

  • The future of AI in science communication and the importance of keeping "humans in the loop"

What the research tells us so far:

  • AI adoption is growing rapidly but unevenly:

    • Nearly half the population in many countries is now using AI tools like ChatGPT

    • Usage for science-related information varies significantly by country (e.g., 21% in Taiwan vs. 4% in Australia)

    • Early adopters tend to be younger, male, more educated, and have positive attitudes toward technology

  • Digital divides are emerging:

    • Studies show higher-educated counties in the US search for AI-related information much more frequently

    • One potential concern is that this could create or widen knowledge gaps as those with AI literacy benefit more from these tools

  • AI conceptualises science in specific ways:

    • When answering scientific questions, AI models generally present a natural science, positivist view of science

    • ChatGPT gives scientifically accurate answers about topics like climate change, regardless of user skepticism

    • However, it adapts its framing based on the user's stance while maintaining scientific accuracy

  • Potential for effective science dialogue:

    • Studies show AI can be effective at reducing belief in conspiracy theories (approximately 20% reduction)

    • Effects appear to be durable, with reduced belief persisting for months after AI conversations

    • AI responses to health questions are often rated higher than physicians' responses for quality and empathy (when users don't know they're from AI)

    • However, when people know that content is AI-generated, they tend to rate it lower

Tools mentioned:

  • Claude and ChatGPT — General large language models

  • Mistral and Gemini - Alternative large language models

  • Elicit and Consensus - Tools that interface with scientific databases

  • JECT AI - Tool for journalists to adapt content for different personas

  • Notebook LM - Google's AI tool that can create audio content

Practical tips from this episode:

  • Start exploring if you haven’t already:

    • AI tools are becoming essential for researchers. Begin experimenting to understand their capabilities and limitations.

    • "Be curious about that. Play around, try to use it... Find your use or see if it has a use for you and my strong assumption would be for many people it actually can be useful to a degree with the human in the loop." -- Mike Schäfer

    • "I think dealing with AI will become a foundational skill and cultural technique that people have to master." -- Mike Schäfer

  • Use AI as a sounding board:

    • Try asking AI for fresh perspectives on how to present your research to specific audiences

    • Use AI to help break out of established patterns when preparing presentations or content

    • "I often have questions like, well, look, I have to give a keynote on this and that. What do you think would be the key messages for such a keynote for this and that audience? And it's always inspiring in a way." -- Mike Schäfer

    • "Very rarely does something come out that I can use as such. But very often something comes out where I can at least take one or two ideas from. So I'm often operating in this gray area of, it doesn't really solve my problems, but it's very, very helpful as an assistant." -- Mike Schäfer

  • Personalise your AI assistant:

    • Create a personalised chatbot by feeding it your publications and professional work

    • Consider how personalised AI tools might help extend your reach and availability

    • "I have personalised my chatbot in a way so it knows who I am and I fed it a number of my professional work so it is familiar with that and what I'm interested in." -- Mike Schäfer

    • "I know colleagues also from the field, like practitioners in science communication that use these chatbots personalised to be trained to be as much like the real thing, the real person as possible." -- Mike Schäfer

  • Try different formats and angles:

    • Ask AI to adapt your content into different formats, styles, or voices to see what might be effective

    • Experiment with visualisations, games, or creative formats that could engage different audiences

    • "I sometimes try to bring in other forms of delivery. So I think about either different voices. Can you do this in a poem? Can you do this like in the, if I don't know, do it Hemingway style or do it like Wired Magazine would write this... Can you visualise that for me? Or can you, if that was a quiz or if that was a game, how would that look like?" -- Mike Schäfer

  • Remember AI is changing both science and communication:

    • Consider how AI is transforming both how we do research and how we communicate about it

    • Stay aware of developments in AI tools specifically designed for science communication

    • "When people think about AI and science communication, they often think about how AI may change the communication side of science communication, which is important, but it will also change the science side of science communication… AI is something that is changing also science profoundly, so many researchers are playing around with science already." -- Mike Schäfer

Credits:

  • Host & Producer: Chris Pahlow

  • Edited by: Laura Carolina Corrigan

  • Music by: La Boucle and Blue Steel, courtesy of Epidemic Sound

Chris Pahlow
Chris Pahlow is an independent writer/director currently in post-production on his debut feature film PLAY IT SAFE. Chris has been fascinated with storytelling since he first earned his pen license and he’s spent the last ten years bringing stories to life through music videos, documentaries, and short films.
http://www.chrispahlow.com
Next

32. Making Sci Comm Training Stick: Brendon Bosworth on finding the right training approach for your team