Capitol News

In a recent study, David Markowitz, an associate professor of communication at Michigan State University, explores how artificial intelligence (AI) can simplify scientific communication and potentially restore public trust in science. Markowitz's research highlights the challenges faced by the general public in understanding complex scientific jargon, which may contribute to declining trust in science.

The study, published in PNAS Nexus, investigates the effectiveness of AI-generated summaries compared to those written by humans. Markowitz used GPT-4 by OpenAI to create simpler versions of scientific papers. These AI-generated summaries employed more common language and were easier to read according to readability indexes.

Markowitz's findings suggest that AI-generated summaries improve public comprehension of scientific studies. Participants who read these simplified statements demonstrated a better understanding of the content and provided more accurate summaries than those who read human-written versions.

Additionally, the study examined perceptions of scientists based on the complexity of their work descriptions. Participants rated scientists whose work was presented in simpler terms as more credible and trustworthy than those described using complex language. Interestingly, participants often misattributed authorship, believing complex texts were AI-generated while assuming simpler ones were human-written.

The role of AI in science communication is expected to grow as its use becomes more accepted within academic publishing. However, ethical considerations must be addressed. Simplifying scientific content with AI might remove important nuances or lead to misunderstandings if not carefully managed. Transparency about AI usage is essential to prevent biases.

Markowitz concludes that while AI tools can aid in simplifying science communication, scientists themselves can achieve similar results by reducing jargon and striving for clarity without relying on technology.