Hybrid AI. Combined use of knowledge and data
In the field of Artificial Intelligence (AI), enormous attention is currently being paid to Machine Learning (ML) with neural networks (Deep Learning) and Large Language Models in particular. The reason for this is their impressive results, for example in the automatic generation of text or images, which are essentially based on large amounts of data, the scalability of the models and high computing capacities. However, fundamental problems of large AI models cannot simply be solved by further scaling, not least because of the enormous consumption of resources. Hybrid AI can help with such problems by combining existing knowledge with learning processes. This results in more efficient, robust, explainable and trustworthy AI systems that are subject to less bias and require smaller amounts of data for the learning process. Hybrid AI can therefore be the better choice, especially in applications that require reliable and accurate results with limited resources, such as in the healthcare sector.
The format AI AT A GLANCE explains in an understandable way how hybrid AI works, what potentials and challenges it harbours and presents concrete application examples.