Mixopoietic AI system governance at the tertiary order of reflexive expectation dynamics

The new wave of Artificial Intelligence, in interdependence with highly complex economicand mass media systems, became in recent years the motor of social disorder, leading to acounter-wave of ethical discussion, with little mitigation results. While new regulatoryframeworks lack a global consensus, the risk of imminent totalitarian positions indicates theneed for a contingency approach to deal with the rise of social uncertainties withoutcompromising individual rights. The purpose of this research is to investigate thecontributions of a novel reflexive expectation model to Artificial Intelligence governancecapable of addressing the temporal challenges of contingency, while examiningconcomitantly the conditions and limitations of the model. Reflexive expectation, asunderstood by systems thinker Niklas Luhmann, is an uncomfortable anticipatory ordering ofthe social field that ignites conflict before it is necessary to create the conditions formanagement and stabilisation through the double contingency between societal systems. Toexplore the contribution of the reflexive expectation model, 26 semi-structured interviewshave been collected from 21 North American and European experts and 5 Generative AImodels. The primary data of human experts have been double qualitatively analysed throughautopoietic reflexivity and thematic analysis, which themes again have been double reviewedby using the outputs from the Generative AI models. Results revealed a tertiary order ofobservation, an epistemological tool that provides the mechanisms to temporarilydifferentiate meaning in before/after to adapt the AI system to societal environments withoutcompromising its conditions for innovation. Multiple implications emerged from the thirdorderof observation, including a temporal meaning for AI system called Mixopoietic AI,formulated from the paradox of allopoietic/autopoietic system and the continuousmicrostructural coupling between AI and social systems. The reflexive expectation modelresulting from those premises is a semi-autonomous organisational systems methodologymade by 8 processes that anticipate AI dysfunctionalities, scrutinise the functional deviationsthrough binary coding, and plan decision programs, personnel and channels to synchronisethe biased codes through temporal meaning differentiation, ultimately mitigating irritabilitiesbetween societal and psychic systems by assessing social defence anxieties. Apart from usingthe lens of Luhmann’s social systems theory, the study incorporates contributions fromMerton’s functional analysis and Menzies-Lyth’s social defence theories. The practicalcontributions to systems management include the non-linear temporal synchronisationbetween AI innovation and social order expectations through the continuous analysis ofsystem/environment differentiation and the anticipation of general/specific strategies forcreating the irritation/fulfilment conditions for AI and societal systems co-evolution.Limitations have been noted in assessing the psychic system and signifying systematically thenew codes for the evolutionary path. Further interdisciplinary studies involvingmetaPsychodynamics, cybersemiotics, and empirical evidence from case studies andexperiments have been recommended.