Attitudes and Perceptions of Generative Artificial Intelligence Chatbots in the Scientific Process of Traditional, Complementary, and Integrative Medicine Research: A Large-Scale, International Cross-Sectional Survey

This large-scale international survey reveals that while a majority of Traditional, Complementary, and Integrative Medicine (TCIM) researchers recognize the potential of Generative AI chatbots to enhance research efficiency and reduce workload, they also identify significant challenges regarding bias and errors, highlighting a critical need for institutional training programs to support their effective integration.

The Gist

Imagine the world of scientific research as a massive, bustling library where thousands of scholars are trying to write the ultimate encyclopedia of human health. Now, imagine a new, incredibly fast, and knowledgeable robot assistant (Generative AI, or GenAI) has just been hired to help these scholars.

This paper is like a giant survey sent out to a specific group of librarians: those who specialize in Traditional, Complementary, and Integrative Medicine (TCIM). These are the experts in ancient wisdom, herbal remedies, yoga, and holistic healing. The authors wanted to know: How do these specific experts feel about this new robot assistant?

Here is the story of their findings, broken down simply:

1. The Setup: Who is in the room?

The researchers sent a digital invitation to nearly 60,000 TCIM experts around the world. About 716 people replied. Think of this as a town hall meeting where the most interested and engaged citizens showed up. They came from all over the globe, mostly from universities and hospitals, and they represented a mix of ages and backgrounds.

2. The Verdict: "We like the tool, but we're nervous about the rules."

The general mood was a mix of excitement and caution.

• The "Yes, Please!" Crowd: Most of the researchers (about 58%) already knew how to use these AI chatbots. They see them as a super-powered calculator for words.

  • The Analogy: Imagine you are writing a novel. The AI is like a tireless ghostwriter who can instantly summarize 1,000 books, translate your story into 20 languages, and fix your grammar.
  • What they love: The biggest benefit? Saving time. 74% said it would reduce their workload. They also love how it helps organize messy data and makes writing papers faster.

• The "Wait a Minute..." Crowd: While they like the speed, they are worried about the quality and honesty of the robot.

  • The Analogy: Imagine the robot ghostwriter is so fast it sometimes makes things up. It might invent a fake quote from a famous doctor or mix up two different herbal remedies.
  • The Fears: The biggest worries were bias (the robot having a hidden opinion), errors (making up facts), and privacy (who is reading our secret notes?). 80% of researchers agreed that these tools raise serious ethical red flags.

3. The Gap: "We want to learn, but no one is teaching us."

This is the most critical part of the story.

• The Desire: Over 70% of the researchers said, "We really want training! Teach us how to use this robot without getting it fired!"
• The Reality: Only about 16% had actually received any formal training. Nearly half said their universities or hospitals don't offer any classes on how to use AI safely.
• The Metaphor: It's like giving everyone a brand-new, high-tech car but handing them the keys without a driving lesson or a map. They are eager to drive, but they are terrified of crashing because no one taught them the rules of the road.

4. How They Actually Use It

The survey revealed a clear pattern in how these researchers use the AI:

• They use it for the "Easy Stuff": They love using it for drafting emails, translating documents, and organizing their thoughts.
• They avoid it for the "Hard Stuff": They are very hesitant to let the AI do the critical thinking, like analyzing complex medical data or reviewing other people's work. They want to keep the "human brain" in charge of the most important decisions.

The Bottom Line

This paper tells us that TCIM researchers are open to the future but cautious about the present. They see Generative AI as a powerful tool that could help them heal the world faster and more efficiently. However, they are screaming for guidance, training, and clear rules.

Without a "driver's ed" course and a "traffic code" for AI, they fear the technology might cause more harm than good. The authors suggest that if we want to use AI to improve holistic medicine, we need to build a bridge between the ancient wisdom of TCIM and the new speed of AI, ensuring that the human element of care and ethics remains at the wheel.

Technical Summary

1. Problem Statement

While Generative Artificial Intelligence (GenAI) chatbots (e.g., ChatGPT, Google Gemini) are rapidly transforming scientific research by enhancing efficiency and accessibility, their integration into Traditional, Complementary, and Integrative Medicine (TCIM) research remains under-explored. TCIM research is distinct due to its reliance on diverse epistemologies, culturally grounded knowledge systems, and complex interdisciplinary methodologies.

• The Gap: There is a lack of empirical data regarding TCIM researchers' attitudes, familiarity, perceived benefits, and ethical concerns regarding GenAI.
• The Risk: Without understanding these specific perspectives, the integration of GenAI risks misappropriating traditional knowledge, failing to address TCIM-specific methodological challenges, and exacerbating issues of bias, transparency, and accountability.

2. Methodology

The study employed a large-scale, international cross-sectional survey design targeting TCIM researchers.

• Sampling Framework:
  • Source: Ovid MEDLINE was searched for articles published between December 15, 2014, and December 15, 2024, covering TCIM domains (e.g., acupuncture, Ayurveda, herbal medicine, yoga).
  • Target Population: Corresponding authors of these articles were identified as the primary TCIM researchers.
  • Exclusion: Non-research articles (editorials, commentaries) were filtered out.
  • Data Extraction: Author names, affiliations, and emails were extracted using R scripts (based on easyPubMed and RedEye packages) and manually verified.
• Survey Administration:
  • Platform: Anonymous online survey via SurveyMonkey.
  • Timeline: August 18, 2025 – September 22, 2025 (5 weeks).
  • Recruitment: 59,045 emails sent; 716 responses received (1.21% overall response rate). 642 respondents met eligibility criteria; 566 completed the survey fully.
• Instrument Design:
  • Sections: Demographics, familiarity with GenAI, usage patterns, perceived benefits, perceived challenges, and institutional training/policies.
  • Analysis: Descriptive statistics (frequencies, percentages) for quantitative data. Thematic content analysis (inductive coding) for open-ended qualitative responses.
• Ethical Compliance: Approved by the University Hospital Tübingen Research Ethics Board. Data is open-access via the Open Science Framework (OSF).

3. Key Contributions

• First International TCIM-Specific Study: This is the first large-scale survey to specifically map the attitudes of TCIM researchers toward GenAI, distinguishing their needs from general biomedical researchers.
• Granular Usage Mapping: The study provides a detailed breakdown of GenAI utility across the scientific lifecycle (literature search, data analysis, manuscript writing, peer review).
• Identification of the "Training Gap": It highlights a critical disconnect between the high demand for GenAI training among researchers and the low availability of institutional support.
• Qualitative Depth: Through thematic analysis, the study uncovers nuanced concerns specific to TCIM, such as the risk of "hallucinations" regarding traditional knowledge and the erosion of critical thinking skills.

4. Key Results

Demographics

• Respondents: 642 eligible participants (mostly male, aged 35–54).
• Geography: Primarily South-East Asia (31.65%), Europe (21.53%), and the Americas (20.23%).
• Roles: Majority were faculty members (52.64%) and senior researchers (65.67%).

Familiarity and Usage

• Familiarity: 58.08% reported familiarity with GenAI chatbots.
• Primary Tool: ChatGPT was the most used tool (46.10% of past usage).
• Current Usage: 18.73% had never used GenAI in a scientific context.
• Future Intent: 76.12% are likely or very likely to use GenAI in future research.
• Perceived Importance: 88.74% believe GenAI will be important or very important to the future of scientific research.

Perceived Benefits (Top Ranked)

1. Workload Reduction: 74.07% agreement.
2. Efficiency in Data Analysis/Experimentation: 71.14% agreement.
3. Accessibility of Scientific Information: 70.37% agreement.
4. Handling Large Datasets: 70.10% agreement.

Usage Patterns by Task

• High Usage: Manuscript writing/editing (36.85% use "always" or "often") and translation of research materials (36.13%).
• Low Usage: Peer review (46.39% "never") and research engagement (44.81% "never").
• Uncertainty: High uncertainty regarding the tool's utility in peer review and data interpretation.

Perceived Challenges

• Ethical/Legal: 80.17% cited concerns regarding transparency, privacy, and consent.
• Bias & Errors: 72.37% worried about biased/skewed data; 66.73% concerned about errors in interpreting scientific terminology.
• Transparency: 72.41% cited limited transparency in decision-making as a major challenge.

Training and Policy

• Institutional Support: Only 16.4% of respondents had received formal training on GenAI.
• Policy Gap: 43.67% reported their institutions offered no guidance or training.
• Demand: 72.07% expressed interest in receiving training; 84.08% felt some or a lot of additional training was necessary.

5. Significance and Implications

• Policy Development: The findings suggest an urgent need for TCIM institutions and journals to develop clear guidelines and standardized training programs to ensure the ethical and rigorous use of GenAI.
• Research Integrity: Addressing the "hallucination" risk (e.g., fabricated citations) and bias is critical to maintaining scientific rigour, particularly in TCIM where traditional knowledge must be accurately represented.
• Equitable Advancement: GenAI has the potential to bridge resource gaps for TCIM researchers in developing regions, provided that training and access are democratized.
• Future Research: The study calls for multilingual survey tools to include non-English speaking researchers, who are often the custodians of non-biomedical traditional knowledge systems, to ensure a truly global perspective.

Conclusion: While TCIM researchers view GenAI as a powerful tool for efficiency and accessibility, significant barriers regarding training, institutional policy, and ethical concerns remain. The responsible integration of GenAI in TCIM requires a collaborative approach to develop tools and policies that respect the unique epistemological and cultural foundations of the field.