This pilot survey explores the adoption, applications, and perceptions of Artificial Intelligence (AI) among Bangladeshi doctors in medical learning, research, and clinical practice. It reveals that AI, primarily ChatGPT, is increasingly used for educational and research purposes like literature review and protocol writing. While doctors recognize AI's benefits for improving these areas, its routine use in clinical practice is less frequent, and most view AI as an augmenting tool rather than a replacement for physicians. The study highlights the need for cautious integration, training, and ethical considerations for AI in healthcare.
Introduction Artificial intelligence (AI) is transforming healthcare across education, research, diagnostics, and clinical decision support. Recent advancements in machine learning and large language models like ChatGPT have accelerated AI adoption among healthcare professionals, offering rapid information retrieval and data analysis. In medical education, AI supports personalized learning and concept understanding, making AI literacy crucial. For research, AI aids in literature review, planning, data analysis, and scientific writing, though requiring human oversight for accuracy and ethical use. Clinical applications show promise in diagnostics and decision support, but concerns about bias, explainability, and accountability remain. This pilot study aims to understand AI use patterns, platforms, applications, and perceptions among Bangladeshi doctors to inform training and integration policies. Materials and methods This was a cross-sectional, internet-based pilot survey conducted among 58 Bangladeshi doctors. A structured questionnaire, developed using Google Forms, was distributed via a WhatsApp group to gather data on demographics, AI use frequency, preferred platforms, applications in learning, research, and clinical practice, and perceptions regarding AI's future role. Participation was voluntary and anonymous, indicating consent for data use. Data were analyzed using descriptive statistics, presenting categorical variables as frequencies and percentages. Formal Institutional Review Board (IRB) approval was not obtained as the study was deemed exempt for involving minimal risk and no identifiable participant information, according to the Bangladesh Medical Research Council guidelines. Results The survey included 58 doctors, with a slight majority being female (51.7%) and most (62.0%) aged 25-34 years. Postgraduate trainees constituted the largest professional group (72.4%). ChatGPT was the dominant AI platform, used by 93.1% of respondents. AI was most frequently used in medical learning, followed by research, and least in clinical practice. Common applications included casual reading (62.1%) for learning, literature review (62.1%) and protocol writing (58.6%) for research, and symptom analysis (60.3%) and diagnostic support (53.4%) for clinical practice. Participants generally perceived a moderate to substantial positive impact of AI, especially in medical learning. Most doctors believed their performance would not significantly decrease without AI (60.3% reported less than 50% decrease, 32.8% no decrease). A majority (55.2%) supported early integration of AI into medical education and healthcare, and while views varied on AI replacing conventional medical activities, a strong majority (82.8%) did not believe AI would replace doctors. Demographic characteristics Out of 58 doctors surveyed, 51.7% were female and 48.3% male. The largest age group was 25-34 years (62.0%), followed by 35-44 years (32.8%). Most participants were postgraduate trainees (72.4%), with specialists making up 25.9% and FCPS Final part examinees 1.7%. No interns or general practitioners participated. Patterns of artificial intelligence platform use ChatGPT was the most widely used AI platform, reported by 93.1% of the 58 respondents. Other platforms used included Claude (12.1%), Gemini (8.6%), Notebook LM (6.9%), GrokAI (3.4%), and SciSpace (3.4%). A few participants also used tools like Perplexity, Copilot, Microsoft 365 AI, and Google AI-based tools. Frequency of artificial intelligence platform use AI use was most frequent in medical learning: 53.4% used it less than 50% of the time, 31.0% more than 50%, and 12.1% all the time. For research, 41.4% used it less than 50% of the time, 36.2% more than 50%, and 12.1% all the time. In clinical practice, usage was lowest, with 67.2% using it less than 50% of the time, 15.5% more than 50%, and only 3.4% all the time. Areas of application of artificial intelligence In medical learning, AI was primarily used for casual reading (62.1%), presentation preparation (41.4%), and exam question preparation (37.9%). For research, common applications included literature review (62.1%), protocol writing (58.6%), topic selection (44.8%), and statistical analysis (44.8%). In clinical practice, AI supported symptom analysis (60.3%), diagnostic support (53.4%), and investigation planning (39.7%). Impact of artificial intelligence Participants perceived AI as having a moderate-to-substantial positive impact across all areas. Moderate improvement (score 3) was the most common response for medical learning (34.5%), research (29.3%), and clinical practice (36.2%). The perceived benefit was strongest in medical learning, followed by research, and then clinical practice. Lack of training for AI in research was cited as a reason for less perceived helpfulness in that domain. Fluctuation of performance without artificial intelligence A majority of participants (60.3%) believed their performance would decrease less than 50% of the time without AI assistance, while 32.8% thought it would never decrease. Only 6.9% felt their performance would decrease more than 50% of the time. This indicates AI is largely seen as an enhancing tool rather than an indispensable one for routine performance. Early integration of artificial intelligence Most respondents (55.2%) agreed that AI should be integrated into medical learning, research, and clinical practice from an early stage. A significant proportion (34.5%) were uncertain, while only 10.3% did not support early integration, reflecting a general openness towards incorporating AI early in medical education and practice. Future role of artificial intelligence: replacement or augmentation? Opinions varied on whether AI would replace conventional medical activities, with 39.7% believing it would not and 32.8% thinking it could. However, a strong majority of participants (82.8%) disagreed with the idea that AI would replace doctors in the future, with only 6.9% believing it would. This suggests a predominant view of AI as an augmentation tool rather than a complete substitute for human medical professionals. Discussion This pilot survey highlights that Bangladeshi doctors are adopting AI, primarily for medical learning and research, with less routine use in clinical practice. ChatGPT's popularity is linked to its accessibility and versatility. While AI aids in educational support, literature review, and protocol writing, challenges like accuracy and ethical concerns persist. Participants view AI as beneficial but emphasize cautious integration, appropriate training, and ethical guidelines. The study suggests AI will likely augment, rather than replace, human physicians, underscoring the enduring importance of human expertise, reasoning, and ethics in healthcare. Limitations include a small sample size, potential selection bias, self-reported data, and lack of questionnaire validation. Conclusions Bangladeshi doctors are increasingly using AI tools, particularly for medical learning and research, with ChatGPT being the most common platform. AI applications include literature review, educational support, symptom analysis, and diagnostic assistance. Doctors perceive AI as beneficial for enhancing these areas, but clinical use remains limited, highlighting the need for structured training, ethical guidance, and responsible integration. The consensus is that AI should serve as a supportive technology to improve efficiency while preserving the critical role of human expertise and judgment, rather than replacing physicians. Further large-scale studies are recommended to thoroughly understand AI adoption patterns and educational needs in Bangladesh.
Introduction
Artificial intelligence (AI) is transforming healthcare across education, research, diagnostics, and clinical decision support. Recent advancements in machine learning and large language models like ChatGPT have accelerated AI adoption among healthcare professionals, offering rapid information retrieval and data analysis. In medical education, AI supports personalized learning and concept understanding, making AI literacy crucial. For research, AI aids in literature review, planning, data analysis, and scientific writing, though requiring human oversight for accuracy and ethical use. Clinical applications show promise in diagnostics and decision support, but concerns about bias, explainability, and accountability remain. This pilot study aims to understand AI use patterns, platforms, applications, and perceptions among Bangladeshi doctors to inform training and integration policies.
Materials and methods
This was a cross-sectional, internet-based pilot survey conducted among 58 Bangladeshi doctors. A structured questionnaire, developed using Google Forms, was distributed via a WhatsApp group to gather data on demographics, AI use frequency, preferred platforms, applications in learning, research, and clinical practice, and perceptions regarding AI's future role. Participation was voluntary and anonymous, indicating consent for data use. Data were analyzed using descriptive statistics, presenting categorical variables as frequencies and percentages. Formal Institutional Review Board (IRB) approval was not obtained as the study was deemed exempt for involving minimal risk and no identifiable participant information, according to the Bangladesh Medical Research Council guidelines.
Results
The survey included 58 doctors, with a slight majority being female (51.7%) and most (62.0%) aged 25-34 years. Postgraduate trainees constituted the largest professional group (72.4%). ChatGPT was the dominant AI platform, used by 93.1% of respondents. AI was most frequently used in medical learning, followed by research, and least in clinical practice. Common applications included casual reading (62.1%) for learning, literature review (62.1%) and protocol writing (58.6%) for research, and symptom analysis (60.3%) and diagnostic support (53.4%) for clinical practice. Participants generally perceived a moderate to substantial positive impact of AI, especially in medical learning. Most doctors believed their performance would not significantly decrease without AI (60.3% reported less than 50% decrease, 32.8% no decrease). A majority (55.2%) supported early integration of AI into medical education and healthcare, and while views varied on AI replacing conventional medical activities, a strong majority (82.8%) did not believe AI would replace doctors.
Demographic characteristics
Out of 58 doctors surveyed, 51.7% were female and 48.3% male. The largest age group was 25-34 years (62.0%), followed by 35-44 years (32.8%). Most participants were postgraduate trainees (72.4%), with specialists making up 25.9% and FCPS Final part examinees 1.7%. No interns or general practitioners participated.
Patterns of artificial intelligence platform use
ChatGPT was the most widely used AI platform, reported by 93.1% of the 58 respondents. Other platforms used included Claude (12.1%), Gemini (8.6%), Notebook LM (6.9%), GrokAI (3.4%), and SciSpace (3.4%). A few participants also used tools like Perplexity, Copilot, Microsoft 365 AI, and Google AI-based tools.
Frequency of artificial intelligence platform use
AI use was most frequent in medical learning: 53.4% used it less than 50% of the time, 31.0% more than 50%, and 12.1% all the time. For research, 41.4% used it less than 50% of the time, 36.2% more than 50%, and 12.1% all the time. In clinical practice, usage was lowest, with 67.2% using it less than 50% of the time, 15.5% more than 50%, and only 3.4% all the time.
Areas of application of artificial intelligence
In medical learning, AI was primarily used for casual reading (62.1%), presentation preparation (41.4%), and exam question preparation (37.9%). For research, common applications included literature review (62.1%), protocol writing (58.6%), topic selection (44.8%), and statistical analysis (44.8%). In clinical practice, AI supported symptom analysis (60.3%), diagnostic support (53.4%), and investigation planning (39.7%).
Impact of artificial intelligence
Participants perceived AI as having a moderate-to-substantial positive impact across all areas. Moderate improvement (score 3) was the most common response for medical learning (34.5%), research (29.3%), and clinical practice (36.2%). The perceived benefit was strongest in medical learning, followed by research, and then clinical practice. Lack of training for AI in research was cited as a reason for less perceived helpfulness in that domain.
Fluctuation of performance without artificial intelligence
A majority of participants (60.3%) believed their performance would decrease less than 50% of the time without AI assistance, while 32.8% thought it would never decrease. Only 6.9% felt their performance would decrease more than 50% of the time. This indicates AI is largely seen as an enhancing tool rather than an indispensable one for routine performance.
Early integration of artificial intelligence
Most respondents (55.2%) agreed that AI should be integrated into medical learning, research, and clinical practice from an early stage. A significant proportion (34.5%) were uncertain, while only 10.3% did not support early integration, reflecting a general openness towards incorporating AI early in medical education and practice.
Future role of artificial intelligence: replacement or augmentation?
Opinions varied on whether AI would replace conventional medical activities, with 39.7% believing it would not and 32.8% thinking it could. However, a strong majority of participants (82.8%) disagreed with the idea that AI would replace doctors in the future, with only 6.9% believing it would. This suggests a predominant view of AI as an augmentation tool rather than a complete substitute for human medical professionals.
Discussion
This pilot survey highlights that Bangladeshi doctors are adopting AI, primarily for medical learning and research, with less routine use in clinical practice. ChatGPT's popularity is linked to its accessibility and versatility. While AI aids in educational support, literature review, and protocol writing, challenges like accuracy and ethical concerns persist. Participants view AI as beneficial but emphasize cautious integration, appropriate training, and ethical guidelines. The study suggests AI will likely augment, rather than replace, human physicians, underscoring the enduring importance of human expertise, reasoning, and ethics in healthcare. Limitations include a small sample size, potential selection bias, self-reported data, and lack of questionnaire validation.
Conclusions
Bangladeshi doctors are increasingly using AI tools, particularly for medical learning and research, with ChatGPT being the most common platform. AI applications include literature review, educational support, symptom analysis, and diagnostic assistance. Doctors perceive AI as beneficial for enhancing these areas, but clinical use remains limited, highlighting the need for structured training, ethical guidance, and responsible integration. The consensus is that AI should serve as a supportive technology to improve efficiency while preserving the critical role of human expertise and judgment, rather than replacing physicians. Further large-scale studies are recommended to thoroughly understand AI adoption patterns and educational needs in Bangladesh.
