As my research process is wrapping up and in its final stages, I have started to reflect on the process as a whole. For the past eight months, I have devoted much of my time to working on my research and answering my research question. I encountered a few challenges that required me to make changes to my methods which changed the original goal of my research. Now that I am nearing the end of this road trip, I can take a moment to reflect on this adventure.

Think back to the initial curiosity that sparked your inquiry. What other curiosities do you have and how has this process prepared you to explore them?

My initial curiosity that sparked my inquiry was artificial intelligence (AI). AI is increasingly being used in different areas such as marketing, banking, transportation, and healthcare. I have always been fascinated by AI, and I think that it’s amazing that this technology can be used to complete tasks and think like humans. However, through my research, I discovered that AI is not perfect, and there is still a lot of work to be done until AI can be largely unbiased. My research process allowed to me explore the implications of AI in healthcare and how this may impact medical systems in developing countries. Another curiosity I have is the use of AI in criminal justice. I briefly studied this when I was analyzing sources for my literature review. I read about an algorithm called COMPAS, in the article “Machine Bias”, which rates the likelihood of a defendant’s risk of committing a future crime. Researchers found that COMPAS labeled black people as high-risk, almost twice as much, as white people (Figure 1, Figure 2). This is a very similar issue to racial bias in AI in healthcare, where an algorithm is more likely to be inaccurate for a person of color due to the lack of diversity in datasets. My research process has prepared me to explore AI in the criminal justice system because I know more background information about racial biases in AI and how they come to be. Additionally, I understand more about the harmful effects that racially biased algorithms have on minorities, and with this knowledge, I can explore deeper into the specific consequences of racially biased algorithms in the criminal justice system. Another curiosity that I have from this research process and is especially relevant today is the use of AI in the COVID-19 pandemic. I read a couple of sources regarding the use of AI in contact tracing, early diagnosis, and the development of vaccines. With my new knowledge about AI in medicine, I hope to explore more about the use of AI in tackling the COVID-19 pandemic as well as its potential in preventing or minimizing future pandemics.

If you could revisit your research process, what would you do differently and why?

If I could revisit my research process I would research my methods more in-depth beforehand. I had only prepared a simplified version of my methods in advance. When I actually implemented the methods I ran into some issues. This caused me to have to make changes to my methods and consequently my research question. At the beginning of my research process, my original research goal was to determine the impact of potential racial biases in artificial intelligence (AI) on the effectiveness of AI to improve healthcare in developing countries. I initially wanted to analyze the representation of each race in medical datasets and determine the difference in false-positive rates for each race in chest x-ray algorithms. I thought that I could do this without the use of an actual algorithm, however, when I actually started my methods I found that I would need an AI algorithm along with test subjects to carry out my research. Given the timeframe as well as a required IRB I could not do this. While I did end up being able to analyze the representation of each race in medical datasets, I was unable to determine the false-positive rates. Accordingly, I adjusted my research goal to determine the impact of racial representation in medical datasets on the effectiveness of artificial intelligence in developing countries.

If I were to revisit my research process I would examine my methods more thoroughly earlier on, going step-by-step and figuring out the logistics. I would make sure that I have the resources I need, and that my methods can be completed in a timely manner. I would regularly check if my methods align with my research question and make modifications to my research question accordingly. Even though changes are almost inevitable in research, by planning out my methods earlier and more in-depth I may have been able to avoid this major alteration to my methods.

References

Angwin, J., Larson, J., Mattu, S., & Kirchner, L. (2016, May 23). Machine Bias. ProPublica. https://www.propublica.org/article/machine-bias-risk-assessments-in-criminal-sentencing

Larson, J., Mattu, S., Kirchner, L., & Angwin, J. (2016, May 23). How We Analyzed the COMPAS Recidivism Algorithm. ProPublica. https://www.propublica.org/article/how-we-analyzed-the-compas-recidivism-algorithm