ADaSci Premium Membership fee will be revised from 1st March 2024. Lock your membership for 1 year at current price.

Analysing the Potential Bias and Fairness of “Artificial Intelligence as a Service”

Author(s): Srinivas Babu Ratnam

Abstract

In the rapidly developing “AI as a Service” (AIaaS) market, plug-and-play versions of artificial intelligence services and tool sets are offered for purchase. Customers (users) of AIaaS have the ability to quickly build and integrate AI capabilities into their applications without the need to create their own systems from the ground up. Despite this, it is well knowledge that AI systems are capable of developing their own forms of bias and inequity. According to the findings of this study, the “one-size-fits-all” nature of AIaaS results in challenges and disagreements due to the fact that fairness varies depending on the circumstances. In this work, we investigate and organize the AIaaS sector, as well as present a taxonomy of artificial intelligence services that is based on the degrees of autonomy that are provided to the client. In the next section, we present a critical examination of the various forms of AIaaS, pointing out how each type may encourage prejudices or create other harm in the context of applications that are intended for the general public. By taking these measures, we hope that academics will start to pay more attention to the challenges that come with this emerging subject.

The Chartered Data Scientist Designation

Achieve the highest distinction in the data science profession.

Elevate Your Team's AI Skills with our Proven Training Programs

Strengthen Critical AI Skills with Trusted Generative AI Training by Association of Data Scientists

Explore more from Association of Data Scientists