This paper attempts to describe and address a specific puzzle related to compositionality in artificial networks such as Deep Neural Networks and machine learning in general. The puzzle identified here touches on a larger debate in Artificial Intelligence related to epistemic opacity but specifically focuses on computational applications of human level linguistic abilities or properties and a special difficulty with relation to these. Thus, the resulting issue is both general and unique. A partial solution is suggested.

In an increasingly complex military operating environment, next generation wargaming platforms can reduce risk, decrease operating costs, and improve overall outcomes. Novel Artificial Intelligence enabled wargaming approaches, based on software platforms with multimodal interaction and visualization capacity, are essential to provide the decision-making flexibility and adaptability required to meet current and emerging realities of warfighting. We highlight three areas of development for future warfighter-machine interfaces: AI-directed decisional guidance, computationally informed decision-making, and realistic representations of decision spaces. Progress in these areas (...) will enable development of effective human-AI collaborative decision-making, to meet the increasing scale and complexity of today’s battlespace. (shrink)

Intersubjectivity refers to one person’s awareness in relation to another person’s awareness. It is key to well-being and human development. From infancy to adulthood, human interactions ceaselessly contribute to the flourishing or impairment of intersubjectivity. In this work, we first describe intersubjectivity as a hallmark of quality dyadic processes. Then, using parent-child relationship as an example, we propose a dyadic active inference model to elucidate an inverse relation between stress and intersubjectivity. We postulate that impaired intersubjectivity is a manifestation of (...) underlying problems of deficient relational benevolence, misattributing another person’s intentions, and neglecting the effects of one’s own actions on the other person. These problems can exacerbate stress due to excessive variational free energy in a person’s active inference engine when that person feels threatened and holds on to his/her invalid beliefs. In support of this dyadic model, we briefly describe relevant neuroimaging literature to elucidate brain networks underlying the effects of an intersubjectivity-oriented parenting intervention on parenting stress. Using the active inference dyadic model, we identified critical interventional strategies necessary to rectify these problems and hereby developed a coding system in reference to these strategies. In a theory-guided quantitative review, we used this coding system to code 35 clinical trials of parenting interventions published between 2016 and 2020, based on PubMed database, to predict their efficacy for reducing parenting stress. The results of this theory-guided analysis corroborated our hypothesis that parenting intervention can effectively reduce parenting stress if the intervention is designed to mitigate the problems of deficient relational benevolence, under-coupling, and over-mentalizing. We integrated our work with several dyadic concepts identified in the literature. Finally, inspired by Arya Nagarjuna’s Buddhist Madhyamaka Philosophy, we described abstract expressions of Dependent Origination as a relational worldview to reflect on the normality, impairment, and rehabilitation of intersubjectivity. (shrink)

As interpersonal, racial, social, and international conflicts intensify in the world, it is important to safeguard the mental health of individuals affected by them. According to a Buddhist notion “if you want others to be happy, practice compassion; if you want to be happy, practice compassion,” compassion practice is an intervention to cultivate conflict-proof well-being. Here, compassion practice refers to a form of concentrated meditation wherein a practitioner attunes to friend, enemy, and someone in between, thinking, “I’m going to help (...) them.” The compassion meditation is based on Buddhist philosophy that mental suffering is rooted in conceptual thoughts that give rise to generic mental images of self and others and subsequent biases to preserve one’s egoism, blocking the ultimate nature of mind. To contextualize compassion meditation scientifically, we adopted a Bayesian active inference framework to incorporate relevant Buddhist concepts, including mind, compassion, aggregates, suffering, reification, conceptual thoughts, and superimposition. In this framework, a person is considered a Bayesian Engine that actively constructs phenomena based on the aggregates of forms, sensations, discriminations, actions, and consciousness. When the person embodies rigid beliefs about self and others’ identities and the resulting ego-preserving bias, the person’s Bayesian Engine malfunctions, failing to use prediction errors to update prior beliefs. To counter this problem, after recognizing the causes of sufferings, a practitioner of the compassion meditation aims to attune to all others equally, friends and enemies alike, suspend identity-based conceptual thoughts, and eventually let go of any identity-grasping belief and ego-preserving bias that obscure reality. We present a brain model for the Bayesian Engine of three components: Relation-Modeling, Reality-Checking, and Conflict-Alarming, which are subserved by the Default-Mode Network, Frontoparietal Network and Ventral Attention Network, and Salience Network, respectively. Upon perceiving conflicts, the strengthening or weakening of ego-preserving bias will critically depend on whether the SN up-regulates the DMN or FPN/VAN, respectively. We propose that compassion meditation can strengthen brain regions that are conducive for suspending prior beliefs and enhancing the attunements to the counterparts in conflicts. (shrink)

In the last 20 years the Turing test has been left further behind by new developments in artificial intelligence. At the same time, however, these developments have revived some key elements of the Turing test: imitation and adversarialness. On the one hand, many generative models, such as generative adversarial networks, build imitators under an adversarial setting that strongly resembles the Turing test. The term “Turing learning” has been used for this kind of setting. On the other hand, AI benchmarks are (...) suffering an adversarial situation too, with a ‘challenge-solve-and-replace’ evaluation dynamics whenever human performance is ‘imitated’. The particular AI community rushes to replace the old benchmark by a more challenging benchmark, one for which human performance would still be beyond AI. These two phenomena related to the Turing test are sufficiently distinctive, important and general for a detailed analysis. This is the main goal of this paper. After recognising the abyss that appears beyond superhuman performance, we build on Turing learning to identify two different evaluation schemas: Turing testing and adversarial testing. We revisit some of the key questions surrounding the Turing test, such as ‘understanding’, commonsense reasoning and extracting meaning from the world, and explore how the new testing paradigms should work to unmask the limitations of current and future AI. Finally, we discuss how behavioural similarity metrics could be used to create taxonomies for artificial and natural intelligence. Both testing schemas should complete a transition in which humans should give way to machines—not only as references to be imitated but also as judges—when pursuing and measuring machine intelligence. (shrink)

Of primary importance in formulating a response to the increasing prevalence and power of artificial intelligence (AI) applications in society are questions of ontology. Questions such as: What “are” these systems? How are they to be regarded? How does an algorithm come to be regarded as an agent? We discuss three factors which hinder discussion and obscure attempts to form a clear ontology of AI: (1) the various and evolving definitions of AI, (2) the tendency for pre-existing technologies to be (...) assimilated and regarded as “normal,” and (3) the tendency of human beings to anthropomorphize. This list is not intended as exhaustive, nor is it seen to preclude entirely a clear ontology, however, these challenges are a necessary set of topics for consideration. Each of these factors is seen to present a 'moving target' for discussion, which poses a challenge for both technical specialists and non-practitioners of AI systems development (e.g., philosophers and theologians) to speak meaningfully given that the corpus of AI structures and capabilities evolves at a rapid pace. Finally, we present avenues for moving forward, including opportunities for collaborative synthesis for scholars in philosophy and science. (shrink)