Abstract
Since the advent of artificial intelligence (AI) and the broader discipline of computer science (CS), the solutions that developers have produced have transformed many aspects of society. The healthcare field in particular benefits from the efficiency at which AI can learn from and use patient data to make more accurate and personalized diagnoses for patients. The STS research explores reasons for distrust and skepticism surrounding AI in healthcare which could slow the mainstream adoption of these beneficial technologies using the Social Construction of Technology model. One of the largest roadblocks that prevents the relevant social groups from trusting in these systems is the low level of explainability of the systems. The technical research report proposes a solution to this problem in the form of an introductory computer science course meant for people with no computer science knowledge. The STS research paper and technical report work together to address and hopefully help to overcome some of the skepticism surrounding the adoption of AI systems in healthcare and other fields. By taking practical steps towards addressing the obstacles, the relevant social groups can help ensure that AI develops and is used in alignment with the interests of society.
After motivating the need for artificial intelligence solutions in healthcare, the STS research paper uses Pinch and Bijker’s Social Construction of Technology model to examine the use of the technology. Once the relevant social groups are identified, the research explores the different obstacles that surround the use of AI in healthcare and answers the question of what practical solutions are available to address such obstacles. The problems and solutions are framed through the lens of their relevance to the social groups. The obstacles that the research identifies are low explainability, bias, confidentiality concerns, and lack of organizational readiness on the part of healthcare systems. The research cites surveys and interviews with medical professionals and the general public in order to prove that these issues are pervasive and require solutions. Methods for preventing the effectiveness of these obstacles are policymaking and lawmaking, engaging and educating professionals in the medical field, and making information on the use of these systems available to the public.
In order to begin addressing the problem of explainability in computer science, the first step is education. There should be an accessible way for students, patients, doctors, and other professionals to educate themselves on the basics of CS without having to enroll in a programming course or participate in any extensive curriculum. The technical project approached this problem by creating an introductory CS course which gives a high level overview of computing and how it is used to solve interdisciplinary problems in our world. The course draws from and dilutes many of the concepts taught in high level computer science courses offered at UVa including Theory of Computation, Algorithms, Artificial Intelligence, Databases, Web Programming Languages, Advanced Software Development, Computer Architecture, and Introduction to Cybersecurity. After participating in this course, students would have a better understanding of computing and how it is applied in society, which would hopefully lead to a higher level of trust for systems that use CS and AI concepts.
The result of the technical project is an introductory computer science course, including a full course website. The website contains a full syllabus, a course schedule, and an example take home assignment that could be used to reinforce one of the concepts. Included in the course schedule are outlines for the topics that will be covered in each class lecture, three full sets of lecture slides, and readings that would supplement the concepts taught in each lecture. The course’s topics provide a strong foundation for any student who wants to move forward in computer science education as well as anyone who is interested in becoming a computer science savvy professional in their discipline.
The STS research suggests that one way to avoid some of the skepticism surrounding AI in medicine is to engage and educate the medical professionals who will be using the systems in patient care. In conjunction with this suggestion, the technical research identifies one possible option for carrying out this education. The overall goal of both the technical and STS portions is to further the dissemination of ethically sound artificially intelligent systems into the healthcare system with the belief that they have the potential to make a positive impact.
