Cloud-based AI platforms

Opportunities and limitations of services for Machine Learning as a Service

In times of advancing digitalization, small and medium-sized companies are faced with the challenge of how artificial intelligence methods can increase value creation in the company. However, competition on the labour market for specialized experts is fierce and the implementation of corresponding projects is expensive. As a result, many AI ideas are threatened with early dissolution, despite possibly great potential.

Cloud-based AI platforms from well-known companies such as Microsoft, Google, IBM and Amazon promise to help. The basic idea is to provide common methods and tools for an AI project intuitively and to automate the work processes of an AI developer in the cloud. On the one hand, this should eliminate the need to purchase expensive hardware and, on the other, enable the platform to be used by non-specialists.

The large number of services and the specificity of the solutions make it difficult to make a selection, especially for those outside the field. There is a lack of overview and evaluation criteria that make it possible to compare AI platforms independently of providers, but adapted to the individual problem. This study derives recommendations for positioning in the machine-learning-as-a-service environment based on several AI use cases relevant to companies.