Is the rule-based order in AI dead, or is still kicking? [Conference paper]

This article focuses on the possibility of revival of the classical rule-based approach in AI for building heterogeneous AI systems. It is based on personal experience while working on several research and innovation projects in diverse areas - business process management, unauthorized intrusion detection, malicious interference protection, digital forensics and diagnostics, data management, etc. Despite their differences, all these areas share something in common: they require multiple operations to be executed in a single transaction and incorporate heuristic rules for different purposes, related to data, knowledge, and operation management. The position of this article is that combining the classical rule-based approach from the early days of AI with more recent developments in AI, such as data-focused machine learning and utility-based reinforcement learning, as well as utilization of the recent technological developments on the cloud and at data centers can be beneficial for widening the real-world application of AI. The challenges, which this complexity creates require joined efforts of academic researchers, industrial engineers, and business enablers. Collaboration between them across the board can be highly beneficial and the author is looking for opportunities in this direction.