Have you ever read an article about a scientific discovery and felt like it was written in a foreign language? If you’re like most Americans, new scientific information can be difficult to understand -; especially if you are trying to tackle a scientific article in a research journal.
At a time when scientific literacy is crucial for informed decision-making, the ability to communicate and understand complex content is more important than ever. Trust in science has been declining for years, and one contributing factor may be difficulty understanding scientific jargon.
New research from David Markowitz, associate professor of communications at Michigan State University, points to a potential solution: using artificial intelligence, or AI, to simplify science communication. His work demonstrates that AI-generated summaries can help restore trust in scientists and, in turn, encourage greater public engagement in scientific issues -; simply by making scientific content more accessible. The issue of trust is particularly important because people often rely on science to inform decisions in their daily lives, from choosing what foods to eat to making crucial health care choices.
The responses are excerpts from an article originally published in The Conversation.
How have simpler AI-generated summaries affected the general public’s understanding of scientific studies?
Artificial intelligence can generate summaries of scientific articles that make complex information more understandable to the public compared to summaries written by humans, according to Markowitz’s recent study, published in PNAS Nexus. AI-generated summaries have not only improved the public’s understanding of science, but also improved the way people perceive scientists.
Markowitz used a large popular language model, OpenAI’s GPT-4, to create simple summaries of scientific articles; this type of text is often called a statement of meaning. The AI-generated summaries used simpler language – ; they were easier to read on a readability index and used more common words, such as “work” instead of “occupation” -; as the summaries written by the researchers who carried out the work.
In one experiment, he found that readers of AI-generated statements had a better understanding of the science and provided more detailed and accurate summaries of content than readers of human-written statements.
How have simpler AI-generated summaries affected the general public’s perception of scientists?
In another experiment, participants rated scientists whose work was described in simple terms as more credible and trustworthy than scientists whose work was described in more complex terms.
In both experiments, participants did not know who wrote each summary. The simplest texts were always generated by AI and the complex texts were always generated by humans. When I asked participants who they thought wrote each summary, they ironically thought that the most complex ones were written by AI and the simplest ones were written by humans.
What do we still need to learn about AI and science communication?
As AI continues to evolve, its role in scientific communication could expand, especially if the use of generative AI becomes more common or sanctioned by journals. Indeed, the field of academic publishing is still setting standards regarding the use of AI. By simplifying scientific writing, AI could contribute to increased engagement on complex issues.
While the benefits of AI-generated science communication may be obvious, ethical considerations must also be taken into account. There is some risk that relying on AI to simplify scientific content could eliminate nuance, potentially leading to misunderstandings or oversimplifications. There is also always a risk of mistakes if no one pays attention. Additionally, transparency is essential. Readers should be informed when AI is used to generate summaries to avoid possible bias.
Simple scientific descriptions are preferable and more beneficial than complex descriptions, and AI tools can help. But scientists could also achieve the same goals by working harder to minimize jargon and communicate clearly -; no AI needed.
Michigan State University has been advancing the common good with uncommon resolve for more than 165 years. One of the world’s leading public research universities, MSU pushes the boundaries of discovery to create a better, safer, and healthier world for all, while providing life-changing opportunities to a diverse and inclusive academic community at across more than 400 study programs in 17 degrees. granting colleges.
Source:
Journal reference:
Markowitz, D. M. (2024). From complexity to clarity: how AI improves scientists’ perceptions and the public’s understanding of science. Nexus PNAS. doi.org/10.1093/pnasnexus/pgae387.