Generative Artificial Intelligence: A rapid review of the attributes that influence its decision making in complex health services (Preprint)

Abstract

<sec> <title>BACKGROUND</title> The advent of Generative Artificial Intelligence (Generative AI or GAI) marks a significant inflection point in AI development. Long viewed as the epitome of reasoning and logic, AI incorporates programming rules that are normative. However, it also has a descriptive component that is based on its programmers’ subjective preferences, and based on any discrepancies in the underlying data. Generative AI generates both truth and falsehood, supports both ethical and unethical decisions, and is neither transparent nor accountable. These factors pose clear risks to optimal decision making in complex health services such as health policy and health regulation. It is important to examine how AI makes decisions both from a rational, normative perspective, and from a descriptive point of view to ensure an ethical approach to AI design, engineering, and use. Therefore, this rapid review was undertaken to identify and map the attributes that influence Generative AI decision making in complex health services. </sec> <sec> <title>OBJECTIVE</title> The objective is to identify and map attributes reported in the literature that influence Generative AI decision making in complex health services. </sec> <sec> <title>METHODS</title> This review was designed to answer the following research question: What attributes have been reported in the literature that influence Generative AI decision making in complex health services? As this topic of interest is evolving rapidly, it merited a rigorous review conducted within a short time frame. Therefore, a rapid review of literature was appropriate. The review provides a clear, reproducible methodology that is reported in accordance with a recognised framework and PRISMA 2020 standards adapted for a rapid review. Inclusion and exclusion criteria were developed, and a database search undertaken within four search systems – ProQuest, Scopus, Web of Science, and Google Scholar. </sec> <sec> <title>RESULTS</title> The results include articles published in 2023 and early 2024. A total of 1,550 articles were identified. After removing duplicates, 1,532 articles remained. Of these, 1,511 articles were excluded based on inclusion and exclusion criteria. Thus, a total of 21 articles were selected. These articles were reviewed in detail to identify Generative AI attributes, count the frequency of the attributes mentioned, and identify attribute types. </sec> <sec> <title>CONCLUSIONS</title> Generative AI brings the promise of advanced automation, but carries significant risks. The results indicate that while Generative AI has attributes that help optimal decision making in complex health services. It also has attributes that are detrimental to this. For example, learning and pattern recognition are helpful, but the lack of a moral compass, lack of empathy, lack of consideration for privacy, and a propensity for bias and hallucination, are detrimental to good decision making. These results suggest that Generative AI programming is neither underpinned by human values nor guided by what ultimately makes us human. </sec> <sec> <title>CLINICALTRIAL</title> - </sec> <sec> <title>INTERNATIONAL REGISTERED REPORT</title> RR2-10.2196/42353 </sec>

Ähnliche Arbeiten