AI-enabled Decision Aid and Decision Support for Symbiotic Autonomous Systems

Symbiotic autonomous systems (SAS) are meant as advanced intelligent information systems which involve a synergistic cooperation and collaboration between a part of broadly perceived autonomous systems and a cognition based part, related to the human stakeholder, and the two parts should operate in a symbiosis. This combination should guarantee an added value in terms of the effectiveness and efficiency of the systems' operation and, above all, problem solving in which a considerable part is related to decision making, also profiting from some added values implied by, for instance, collective intelligence and behavior. A growing complexity of the world implies a growing complication of decision making problems, for instance, related to more and more stakeholders, much uncertain, imprecise and lacking information, and an explicit human centricity, that is, a crucial role of the human being in all decision processes. This all complicated these processes in particular in the context of the SASs. We are concerned with an effective and efficient setting, at least now and the near future, to run such decision processes in which the human being is still a decisive stakeholder but he/she should be aided or supported by some additional “units”, humans (e.g. advisors) or “machines” (e.g. algorithms and computer systems); the human being should not be replaced, at least for dealing with sophisticated tasks. We first consider the case when a human decision maker, who knows about the domain of his activities but not necessarily about solution tools, is aided by a domain expert who knows solution tools but not necessarily the domain. We advocate the use of the judge-advisor type approaches. They are augmented with some AI (artificial intelligence) tools and techniques, notably machine learning models which can help find patterns and relationships in bigger and bigger data sets. Moreover, we deal with the consistency of the AI models employed with some inherent human characteristics, the cognitive biases, and show that this is crucial for a synergistic collaborative partnership between the human being and AI models and systems, notably operating autonomously. We also show how such a synergistic cooperation and collaboration can be implemented by employing a decision support system (DSS) to combine the power of the computer and the power of the humans, notably using the data or document driven DSSs. This gives rise to a so-called AI-enabled DSSs which employ advanced tools and techniques of AI (artificial intelligence) for the gathering and analysis of data and evidence, identification and diagnosis of problems, proposing courses of actions with their analyses, etc. to better “mimic” human cognitive capabilities, observe some inherent human characteristic features, provide comprehensive and trustworthy procedures and forms of results, and - in general - help alleviate limitation of a sheer autonomy of the system. Some examples on supporting decision making in socio-economic planning will be presented.