Have you ever read about a scientific discovery and felt like it was written in a foreign language? If you like most americansit can be difficult to understand new scientific information, especially when trying to work on scientific papers in research journals.
David Markowitz, an associate professor in the Department of Communication at Michigan State University, studies how language influences social and psychological processes.
In an era when scientific literacy is essential for informed decision-making, the ability to communicate and understand complex content is more important than ever. Trust in science is has been in decline for yearsAnd one of the factors may be the challenge of understanding scientific terminology.
new research from david markowitzAn associate professor of communications at Michigan State University points to using artificial intelligence (AI) to simplify science communication as a potential solution. His research shows that AI-generated summaries can help restore trust in, and in turn, encourage, scientists. Expanding public involvement in scientific issues — just makes science content more approachable. Issues of trust are especially important because people often rely on science to make decisions in their daily lives, from choosing the foods they eat to making important health care choices.
The answer is an excerpt from an article originally published on The Conversation.
How have simpler AI-generated summaries impacted the public’s understanding of scientific research?
According to Markowitz’s recent research published in PNAS Nexus, artificial intelligence can generate summaries of scientific papers that make complex information easier for the general public to understand compared to summaries written by humans. AI-generated summaries not only improved the public’s understanding of science, but also strengthened people’s perceptions of scientists.
Markowitz used OpenAI’s GPT-4, a popular large-scale language model, to create simple summaries of scientific papers. This type of text is often called a materiality statement. The AI-generated summaries used simpler language than summaries written by the researchers who performed the work. It was easy to read according to the readability index and used more common words such as “job” instead of “occupation.”
One experiment found that readers of AI-generated statements understood the science better and provided more detailed and accurate summaries of content than readers of human-written statements.
How have simpler AI-generated summaries affected the public’s perception of scientists?
In another experiment, participants rated scientists whose research was explained in simple terms as more trustworthy and trustworthy than scientists whose research was explained in more complex terms.
In both experiments, participants did not know who wrote each summary. Simple text has always been generated by AI, and complex text has always been generated by humans. When participants were asked who they thought wrote each summary, ironically they thought the more complex summaries were written by an AI, while the simpler summaries were written by a human.
What do we still need to learn about AI and science communication?
As AI continues to evolve, its role in science communication is likely to expand, especially if the use of generative AI becomes more common or sanctioned by journals. In fact, norms around the use of AI are still being established in the academic publishing field. By simplifying scientific writing, AI has the potential to make it easier to tackle complex problems.
While the benefits of AI-generated science communication are perhaps obvious, ethical considerations must also be considered. Relying on AI to simplify scientific content runs the risk of losing nuance, which can lead to misunderstandings and oversimplifications. If no one is careful, there is always the possibility of errors. Additionally, transparency is important. To avoid potential bias, readers should be informed if AI is used to generate summaries.
Simple scientific explanations are preferred and more informative than complex explanations, and AI tools can help. However, scientists can also achieve the same goal by minimizing jargon and striving to communicate clearly. No AI required.
