This paper describes and frames a central ethical issue–the liability problem–facing the regulation of artificial computational agents, including artificial intelligence (AI) and robotic systems, as they become increasingly autonomous, and supersede current capabilities. While it frames the issue in legal terms of liability and culpability, these terms are deeply imbued and interconnected with their ethical and moral correlate–responsibility. In order for society to benefit from advances in AI technology, it will be necessary to develop regulatory policies which manage the risk and liability of deploying systems with increasingly autonomous capabilities. However, current approaches to liability have difficulties when it comes to dealing with autonomous artificial agents because their behavior may be unpredictable to those who create and deploy them, and they will not be proper legal or moral agents. This problem is the motivation for a research project that will explore the fundamental concepts of autonomy, agency and liability; clarify the different varieties of agency that artificial systems might realize, including causal, legal and moral; and the illuminate the relationships between these. The paper will frame the problem of liability in autonomous agents, sketch out its relation to fundamental concepts in human legal and moral agency–including autonomy, agency, causation, intention, responsibility and culpability–and their applicability or inapplicability to autonomous artificial agents.

Team decision problems are one of the fundamental problems in decentralized decision making. Because team decision problems are NP-hard, it is important to find methods for reducing their complexity. Wu and Lall gave a definition of sufficient statisticsfor team decision problems, and demonstrated that these statistics are sufficient for optimality, and possess other desirable properties such as being readily updated when additional information becomes available. More recently, Lemon and Lall defined weak sufficient statistics for team decision problems, and showed that these statistics are sufficient for optimality and necessary for simultaneous optimality with respect to all cost functions. This prior work studied the extent to which the complexity of team decision problems can be reduced while maintaining exact optimality. However,when faced with a computationally difficult problem, we are often willing to sacrifice exact optimality for significant reductions in complexity. In this paper we define approximate sufficient statistics, which are a generalization of weak team sufficient statistics. Then we prove that these statistics are quantifiably close to being optimal.

However, foundational problems remain in the either mindfully or inadvertently by individuals or teams of continuing development of AI for team autonomy, humans. One worry about this bright future is that jobs especially with objective measures able to optimize team may be lost; from Mims (2015), function, performance and composition. Something potentially momentous is happening inside AI approaches often attempt to address autonomy by startups, and it's a practice that many of their established modeling aspects of human decision-making or behavior.

Public goods, like food sharing and social health systems, may prosper when prior agreements to contribute are feasible and all participants commit to do so. Yet, free-riders may exploit such agreements, requiring then committers to decide whether to enact the public good when others do not commit. So deciding removes all benefits from free-riders but also from those who are willing to establish the beneficial resource. Here we discuss our work wherein we show, within the framework of the one-shot Public Goods Game (PGG) and using methods of Evolutionary Game Theory (EGT), that (i) implementing extra measures, delimiting benefits to free-riders, often leads to more favorable societal outcomes, especially in larger groups and highly beneficial public goods situations, even if so doing is costlier, and (ii) when restriction mechanism is not available, participation level (i.e. how many other players commit to the PGG cooperation) plays a crucial role in the decision making of commitment proposers, for their survival as well as for promoting the emergence of cooperation. Hence, there exist ethical fine tunings to be observed whenever establishing PGGs, be they for humans or non-humans, for otherwise the supporting joint moral ground may escape from under everyone’s feet.

We discuss philosophical and ethical issues that arise from a dialogue system intended to portray a real person, using recordings of the person together with a machine agent that selects recordings during a synchronous conversation with a user. System output may count as actions of the speaker if the speaker intends to communicate with users and the outputs represent what the speaker would have chosen to say in context; in such cases the system can justifiably be said to be holding a conversation that is offset in time. The autonomous agent may at times misrepresent the speaker's intentions, and such failures are analogous to good-faith misunderstandings. The user may or may not need to be informed that the speaker is not organically present, depending on the application.

Based on the development of several different HRI scenarios using different robots, we have been establishing the SERA ecosystem. SERA is composed of both a model and tools for integrating an AI agent with a robotic embodiment, in humanrobot interaction scenarios. We present the model, and several of the reusable tools that were developed, namely Thalamus, Skene and Nutty Tracks. Finally we exemplify how such tools and model have been used and integrated in five different HRI scenarios using the NAO, Keepon and EMYS robots. Figure 1: Our methodology as an intersection of CGI animation, Human-robot interaction (HRI) systems are spreading as a IVA and robotics techniques.

When designing a robot to interact with people, the decision to incorporate a robot arm may arise. In this paper, we investigate adding an inexpensive, functional arm to our mobile CoBot service robots. Specifically, we examine two-dimensional extendable pantograph arms for CoBot. Pantograph arms have intuitive kinematics and inverse kinematics. Pantograph arms are modular and adding additional linkages corresponds to simple changes in the kinematic calculations. These arms have several advantages (and disadvantages) compared to traditional robot arms. A prototype pantograph arm is currently in development and our goal is to attach a modular pantograph arm to CoBot to perform simple needed tasks, such as knocking on doors and pressing elevator buttons.

There are many successful methods for transferring information from one agent to another. One approach, taken in this work, is to have one (source) agent demonstrate a policy to a second (target) agent, and then have that second agent improve upon the policy. By allowing the target agent to observe the source agent's demonstrations, rather than relying on other types of direct knowledge transfer like Q-values, rules, or shared representations, we remove the need for the agents to know anything about each other's internal representation or have a shared language. In this work, we introduce a refinement to HAT, an existing transfer learning method, by integrating the target agent's confidence in its representation of the source agent's policy. Results show that a target agent can effectively 1) improve its initial performance relative to learning without transfer (jumpstart) and 2) improve its performance relative to the source agent (total reward). Furthermore, both the jumpstart and total reward are improved with this new refinement, relative to learning without transfer and relative to learning with HAT.

Control of the air in envisioned large-scale battles against near-peer adversaries will require revolutionary new approaches to airborne mission management, where decision authority and platform autonomy are dynamically delegated and functional roles and combat capabilities are assigned across multiple distributed tiers of platforms and human operators. System capabilities range from traditional airborne battle managers, to manned tactical aviators, to autonomous unmanned aerial systems. Due to the overwhelming complexity, human operators will require the assistance of advanced autonomy decision aids with new mechanisms for operator supervision and management of teams of manned and unmanned systems. In this paper we describe a conceptual distributed mission management approach that employs novel human-automation teaming constructs to address the complexity of envisioned operations in highly contested environments. We then discuss a cognitive engineering approach to designing roleand task-tailored human machine interfaces between humans and the autonomous systems. We conclude with a discussion of multi-level evaluation approaches for experimentation.

Datasets with hundreds to tens of thousands features is the new norm. Feature selection constitutes a central problem in machine learning, where the aim is to derive a representative set of features from which to construct a classification (or prediction) model for a specific task. Our experimental study involves microarray gene expression datasets; these are high-dimensional and noisy datasets that contain genetic data typically used for distinguishing between benign or malicious tissues or classifying different types of cancer. In this paper, we formulate feature selection as a multiagent coordination problem and propose a novel feature selection method using multiagent reinforcement learning. The central idea of the proposed approach is to "assign" a reinforcement learning agent to each feature where each agent learns to control a single feature; we refer to this approach as MARL. Applying this to microarray datasets creates an enormous multiagent coordination problem between thousands of learning agents. To address the scalability challenge we apply a form of reward shaping called CLEAN rewards. We compare in total nine feature selection methods, including state-of-the-art methods, and show that the proposed method using CLEAN rewards can significantly scale-up, thus outperforming the rest of learning-based methods. We further show that a hybrid variant of MARL achieves the best overall performance.