Drivers and Inhibitors for Organizations’ Intention to Adopt Artificial Intelligence as a Service

Abstract

The adoption of artificial intelligence promises tremendous economic benefits for organizations. Yet, many organizations struggle to unlock the full potential of this technology. To ease the adoption of artificial intelligence for organizations, several cloud providers have begun offering artificial intelligence as a service (AIaaS). Extant research on AIaaS exhibits a strong focus on technical aspects and has opposing views on what drives or inhibits the adoption of AIaaS within organizations. In this research, we synthesize extant research on AIaaS adoption factors and conduct semi-structured interviews with practitioners. Our research yields 12 factors that drive and another 12 factors that inhibit the adoption of AIaaS in practice. We thereby close a gap in scholarly knowledge on adopting this emerging service technology, especially on inhibiting factors, and help guide future research on related behavioral and technical aspects.

Artificial Intelligence as a Service

“…Following the market trends of AIaaS, researchers in diverse disciplines, including information systems, computer science, and management, have started to focus their research on provisioning AI capabilities from the cloud. Diverse research streams on AIaaS recently emerged that deal, for example, with the design and evaluation of AI services (Boag et al 2018; e.g., Elshawi et al 2018), the adoption and effective use of AIaaS (e.g., Zapadka et al 2020;Pandl et al 2021), uncovering AIaaS misuse by its users (e.g., Javadi et al 2020), or understanding AIaaS's issues and vulnerabilities (e.g., Truex et al 2019).…”

“…Different types of inference as a service are accessible on-demand nowadays, such as language services (e.g., text analytics or translation), analytics services (e.g., product recommendations or knowledge inference from big data), speech services (e.g., text-to-speech, speech-to-text), or computer vision services (e.g., analyzing of images and videos in order to find and identify objects, text, and labels) (Javadi et al 2020;Pandl et al 2021). It is easy for developers of all skill levels to use machine learning technology by relying on pre-trained models (Ramesh 2017).…”

“…Consequently, large, high-quality data sets are primarily found in data silos of a few large enterprises, and in contrast, there are only a few freely available high-quality data sets, which are limited to a handful of specific application areas (Thiebes et al 2020). Cloud-based AI infrastructure services may counteract these issues by enabling efficient data storage and sharing for a large amount of AI data and respective models (Pandl et al 2021). Such data can be used internally to train AI models, and also be provided by data as a service providers on users' request via data APIs or web interfaces with granular authentication and authorization controls and pricing models (e.g., volume-based or data type-based subscriptions) (Javadi et al 2020).…”

“…Best practices, cost estimations, and a suitable amount and quality of training data for developing own AI applications are often not readily available for SMEs. In response to this, AIaaS simplifies the usage of AI technologies and makes AI accessible through complexity abstraction (Pandl et al 2021). Complexity abstraction not only relates to hiding implementation details of an AIaaS and its underlying computing layers but also to handing over the control and responsibility of the service to the AIaaS provider.…”

“…AIaaS also achieves high degrees of automation because AIaaS enables users to optimize their AI models automatically, provides a selection of the most suitable hardware architectures, and handles hard-and software failures in an automated manner (Zapadka et al 2020;Pandl et al 2021). Thereby, automation impacts each AIaaS stack layer.…”