Sunshine for scientists: AI can assist but not replace them, University of Florida research finds

University of Florida researchers have conducted a comprehensive study examining whether generative AI can replace human scientists in academic research, finding that while AI excels at certain stages of the research process, it fundamentally falls short in others. This mixed result offers reassurance to research scientists concerned about job displacement while highlighting the emergence of a new “cyborg” approach where humans direct AI assistance rather than being replaced by it.

The big picture: Researchers at the University of Florida tested popular AI models including ChatGPT, Microsoft Copilot, and Google Gemini across six stages of academic research, finding the technology can serve as a valuable assistant but not a replacement for human scientists.

Key findings: AI demonstrated effectiveness in early research stages like ideation and research design but struggled significantly with literature reviews, results analysis, and manuscript production.

• The study, titled “AI and the advent of the cyborg behavioral scientist,” limited human intervention to see how well AI could navigate the entire research process independently.
• Researchers identified that AI requires substantial human oversight in critical analytical areas, functioning more as a tool than a collaborator.

What they’re saying: “A pervasive fear surrounding these AIs is their ability to usurp human labor,” explained Geoff Tomaino, assistant professor in marketing at the University of Florida Warrington College of Business.

• “In general, we found that these AIs can offer some assistance, but their value stops there, as assistance. These tools can do a great deal of legwork. However, the researcher still has a vital place in the process, acting as a director and critic of the AI, not an equal partner.”
• Tomaino also noted the personal dimension: “As these AI tools evolve, it will be up to each individual researcher to decide for which steps of the research process they want to become a cyborg behavioral researcher, and for which they would like to remain simply human.”

Practical implications: The research team advises maintaining high skepticism toward AI outputs, treating them as starting points that require human verification rather than finished products.

• For academic journals, they recommend developing policies that require disclosure of AI assistance and largely prohibit AI use in the peer review process.

Why this matters: As generative AI capabilities expand rapidly, understanding the technology’s genuine limitations in complex knowledge work helps organizations develop more realistic implementation strategies rather than over-investing in capabilities that ultimately require significant human oversight.