We’ve all seen the movies, from The Terminator to I, Robot, to WALL-E, that depict artificial intelligence (AI) as an all-powerful tool that rises against humanity or hinders our growth and development. As generative AI has rapidly evolved in recent years, discussions about its uses and effects have become increasingly divided. Some herald its potential, while others worry about its effects on cognitive development and health, potential job displacement, misinformation, and environmental impacts. Amid these fears, it is equally important to explore how this AI can enhance our capabilities and streamline work. Dr. Kristin Sainani, a professor of Epidemiology and Population Health, takes an optimistic approach, inviting us to consider AI’s potential to enhance critical and creative thinking in fields like clinical research.
In collaboration with fellow statistician, professor, and science journalist, Dr. Regina Nuzzo, Sainani has investigated how AI could be harnessed in clinical research practices. At a recent conference, Sainani recounted that a speaker claimed AI would reduce people’s ability to think critically, to which she responded, “I think it is just the opposite. AI is actually going to make critical thinking all that matters by replacing a lot of the rote and algorithmic tasks.” By automating routine tasks, she argues that it frees researchers to focus on more complex analyses and data visualization. “Think of AI as a starting point,” she emphasized, asserting that when researchers like her and Nuzzo leverage tools like ChatGPT, they boost their productivity and efficiency.
Sainani and Nuzzo shared several use cases highlighting AI’s clear research benefits, most notably its capacity for direct data analysis. After uploading raw datasets, researchers gain immediate, actionable insights without needing advanced programming skills. Sainani says this is particularly useful for checking the numbers behind published research papers. Furthermore, AI can kickstart creativity when it comes to visualizing and presenting data. Researchers can discover innovative presentation methods, generate R code for visualizations, and improve clarity and engagement with their findings.
Almost everyone experiences unconscious bias, including researchers. In data analysis, Sainani and Nuzzo point out that subconscious preferences can sometimes influence researchers’ interpretation or manipulation of data. AI can reduce this bias by providing a blinded version of the data, either with adjusted numbers or hidden treatment labels, which helps researchers avoid skewing their results.
Additionally, Sainani and Nuzzo explain that AI can serve as a translator for technical languages, including medical jargon and statistical analyses. For instance, when analyses are poorly articulated, Sainani will drop the text into ChatGPT and ask it to rephrase the findings into a mathematical model or plain English. “[AI] does an amazing job at guessing what authors are trying to say,” she commented. “It does better than I can without a lot of time spent on my part.” This capability is crucial for improving communication and accessibility within the scientific community and beyond.
Concerns about reliance on AI for writing, especially for students, are widespread. However, Sainani and Nuzzo argue that AI is a poor writer but can serve as a valuable collaborator. In brainstorming, researchers can explore ideas, themes, or outlines and receive immediate feedback. AI’s built-in dictation tool enhances this effort by transforming verbal thoughts into coherent text, helping writers at all levels overcome writer’s block and providing a solid starting point. As an editor, the team notes that AI can help writers become more skilled and efficient by providing detailed feedback on common mistakes and overall flow. Reflecting on her experience, Sainani said, “I learned to write by sitting down with an editor and going line by line…[AI] is a great tutor for that. It can go back and forth and it can tell you why or what it’s picking up that is making your writing unclear.” If used correctly, AI becomes a tool to amplify creativity and storytelling rather than a hindrance to critical thinking and writing skill development.
Beyond analysis, translation, and editing, Sainani and Nuzzo propose that AI can serve as an adversarial collaborator. By asking AI to review, critique, and challenge their work, much like a journalist, research colleague, or patient advocate, researchers can uncover weaknesses and biases in their work. This approach, they explain, mirrors traditional peer reviews but offers greater potential for researchers to refine their methodologies and reinforce the rigor of their findings.
Sainani and Nuzzo note that these examples illustrate AI’s vast potential across various fields but can also extend beyond professional uses into personal lives. For instance, Sainani has used AI as a therapeutic outlet to vent frustrations and gain clarity on a situation. She has also used it to create study guides and practice tests for her daughter’s French class. Nuzzo has leveraged AI as a personalized listening coach to help her recognize bird songs through her cochlear implant, showcasing how this technology can enhance accessibility and learning.
However, they acknowledge that the safe and ethical use of AI is still under evaluation. They stress the need for fact-checking and validating AI-generated content, asking the right questions and carefully reviewing and maintaining a critical eye on the outputs. Despite its impressive capabilities, AI should be seen as an extension of human intellect, empowering researchers and the broader population to explore new ideas and enhance their skills, rather than a replacement.
In this continuously evolving landscape, Sainani and Nuzzo’s work exemplifies a fresh take on how we can harness AI to improve efficiency and productivity while boosting creativity and critical thinking in clinical research and beyond. By embracing these technologies responsibly, we can unlock new possibilities for discovery and innovation, ultimately enhancing our understanding and application of research in various fields.
Dr. Kristin Sainani is a Professor (Teaching), Epidemiology and Population Health at Stanford Medicine. In Human Biology, she teaches an introductory course on probability and statistics, HumBio 89, Introduction to Health Sciences Statistics. Dr. Regina Nuzzo is a lecturer in the Department of Primary Care & Population Health at Stanford Medicine, a freelance science writer, and a statistics professor at Gallaudet University. Join Sainani and Nuzzo on their podcast, Normal Curves , where they dissect popular scientific studies on topics with humor and clarity.