AutoML, low- and no-code promise to make AI easier to use, by reducing the programming skills and development effort required. These new approaches therefore also make it easier for SMEs to get started. They thus offer the potential for rapid dissemination of AI solutions. The aim of this study is to investigate whether these promises can be realised.
After a theoretical introduction, the advantages and limitations of these software tools are discussed. Several no-code, low-code and AutoML tools are then analysed. They provide an insight into the relevant market. Subsequently, the structural characteristics of German SMEs with an impact on the implementation of AI is rounded off by the evaluation of an online survey of AI experts with a focus on SMEs.
Low-code and no-code in conjunction with AutoML partially fulfil the promise of simplified access to AI applications: The existing product landscape already offers a wide range of such software solutions. This can reduce the amount of resources that organisations spend on time-consuming, repetitive tasks. In the long term, the introduction of these tools will change the type of expertise required. However, it cannot fully replace the need for Machine Learning (ML) expertise. The use of ML in companies will therefore continue to require AI specialists and thus a corresponding build-up of expertise, especially in SMEs. This human expertise and understanding of data is needed, for example, to identify the right use cases where ML can add measurable value, or to realise the usability of these models in business processes. This knowledge also remains important with regard to data protection and compliance requirements, as well as adherence to ethical standards.
