Gabriela Penelopé Carolus

New researchers encounter various tools as they begin their academic journey. You might have already learned about AI tools. Alternatively, you might be overwhelmed by the volume of AI tools and question their utility. These questions can leave you in the dark about which to select or if you should abide by the rumours or fears of not incorporating another tool into your research project. 

Fear not, simply put, Artificial Intelligence (AI) is changing scientific discovery by automating data analysis, hypothesis testing, and pattern recognition in fields such as astronomy, healthcare, materials science, and bioinformatics. Machine learning (ML) and generative adversarial networks are speeding up research and creating new opportunities for innovation. AI techniques in materials characterisation are improving efficiency and accuracy, potentially addressing reproducibility challenges in scientific research. In doing so, it adds expert knowledge to AI models, improving their ability to predict future scientific discoveries and inventions. Data-driven AI/ML innovations can speed up scientific discoveries by dealing with challenges such as high system complexity, large search space, incomplete knowledge, and small datasets.

Given the numerous advantages, let’s delve into the use of a specific tool-Generative Artificial Intelligence (AI). This tool, increasingly embraced by early career researchers, holds significant value and credibility in education and scientific applications. In this blog, I will share my insights on its use in Public Health research and highlight its potential to advance public health innovations.

Value vs reputation

Artificial Intelligence (AI) in scientific research raises valid concerns, including potential errors, ethical issues, and misconduct. The reputation of AI being negative stems from the fact that there are risks associated with AI-generated content, which has the potential to compromise scientific integrity. Furthermore, AI models may convey biases and false information. Therefore, as an early career researcher, it is crucial to integrate AI responsibly to maximise its potential without compromising scientific rigour. While there are concerns about AI’s impact on research, addressing these challenges alongside the exciting possibilities AI brings to one’s research is essential.

Scientific Use case

The value of this tool lies in its versatility. Scientists apply Artificial Intelligence (AI) across a wide range of disciplines, including chemistry, biology, medicine, engineering, and computer science. In the medical field, AI aids in diagnostics, treatment planning, and outcome prediction. Computer scientists use AI for network intrusion detection and game development. Additionally, engineers integrate AI into defence services, smart homes, and autonomous vehicles. These areas utilise various AI techniques, with artificial neural networks being the most common. Other techniques include fuzzy expert systems, evolutionary algorithms, and hybrid intelligent systems. The widespread integration of AI has enhanced efficiency and quality across diverse domains.

AI use in public health

In my field of interest, Public Health, I discovered that AI applications include spatial modelling, risk prediction, disease forecasting, and health diagnosis. Despite challenges like limited infrastructure and ethical considerations, AI can revolutionise African healthcare and research, particularly in disease surveillance, diagnostics, health communication, knowledge translation, health literacy/education and treatment optimisation. The integration into health informatics holds promise for improving public health outcomes across the continent, with implementation varying across regions due to differences in technology adoption.

Ensuring AI accuracy through scientific validation

AI is a powerful tool, but it’s not without its limitations. Ensuring AI accuracy in public health research involves modernising data governance, addressing workforce skills gaps, and setting up strategic partnerships while considering ethical implications. Combining AI with human expertise can improve diagnostic accuracy, but challenges like algorithmic bias and generalizability issues persist. Robust clinical evaluation and regulatory frameworks are essential for evidence generation and patient safety. Scientific validation is crucial, and caution is needed when using AI for diagnostic purposes. As such, I would caution graduate students to recognise these limitations and urge them to communicate the uncertainties of this tool when using AI systems to make decisions. 

AI Learning in my public health research

I am committed to improving my ability in data-driven AI/ML innovations for public health research. As I move forward, I am aware of the ethical implications of AI applications. My aim is to advocate for transparency, reproducibility, and health equity while contributing to the progression of research methodologies for the development of AI-powered applications.

I am establishing strategic partnerships with experts across various fields, such as AI/ML, human-computer interaction, design science, medicine, and software engineering, while also actively engaging with the public. This effort has been instrumental in helping me recognize the capabilities and limitations of working with an emerging tool. As a burgeoning researcher, I am mindful of both AI’s strengths and weaknesses. I am dedicated to embracing collaborative and transdisciplinary approaches through continuous learning to fully leverage the potential of these tools in academic environments.