A commitment represents an agent's intention to attempt to bring about some state of the world that is desired by some agent (possibly itself) in the future. Thus, by making a commitment, an agent is agreeing to make sequential decisions that it believes can cause the desired state to arise. In general, though, an agent's actions will have uncertain outcomes, and thus reaching the desired state cannot be guaranteed. For such sequential decision settings with uncertainty, therefore, commitments can only be probabilistic. We argue that standard notions of commitment are insufficient for probabilistic commitments, and propose a new semantics that judges commitment fulfillment not in terms of whether the agent achieved the desired state, but rather in terms of whether the agent made sequential decisions that in expectation would have achieved the desired state with (at least) the promised probability. We have devised various algorithms that operationalize our semantics, to capture problem contexts with probabilistic commitments arising because action outcomes are uncertain, as well as arising because an agent might realize over time that it does not want to fulfill the commitment.

We encounter the results of category formation every day, from demographic categories like race and gender, to role-playing-game classes like "fighter" or "mage". Category membership is often not simply based on the possession of discrete properties but instead constructed from and reflect the highly nuanced relationships (gradience) between members and best-example individuals called "prototypes". In this paper, we present The Marginal, an artificial intelligence (AI)-driven game that (1) computationally models the cognitive categories that players develop when customizing videogame avatars and (2) generates challenges for players to use their perception of visual, textual, and numerical data to progress in a game created using these models. We use archetypal analysis, an AI clustering approach for identifying boundary points in data, to generate tasks in The Marginal for its gameplay. It shows how AI can be combined with games to model and evaluate cognitive  categorization phenomena.

External factors are hard to model using a Markovian state in several real-world planning domains. Although planning can be difficult in such domains, it may be possible to exploit long-term dependencies between states of the environment during planning. We introduce weighted state scenarios to model long-term sequences of states, and we use a model based on a Partially Observable Markov Decision Process to reason about scenarios during planning. Experiments show that our model outperforms other methods for decision making in two real-world domains.

Hand-coded finite-state machines and behavior trees are the go-to techniques for artificial intelligence (AI) developers that want full control over their character's bearing. However, manually crafting behaviors for computer-controlled agents is a tedious and parameter-dependent task. From a high-level view, the process of designing agent AI by hand usually starts with the determination of a suitable set of action sequences. Once the AI developer has identified these sequences he merges them into a complete behavior by specifying appropriate transitions between them. Automated techniques, such as learning, tree search and planning, are on the other end of the AI toolset's spectrum. They do not require the manual definition of action sequences and adapt to parameter changes automatically. Yet AI developers are reluctant to incorporate them in games because of their performance footprint and lack of immediate designer control. We propose a method that, given the symbolic definition of a problem domain, can automatically extract a transparent behavior model from Goal-Oriented Action Planning (GOAP). The method first observes the behavior exhibited by GOAP in a Monte-Carlo simulation and then evolves a suitable behavior tree using a genetic algorithm. The generated behavior trees are comprehensible, refinable and as performant as hand-crafted ones.

Recent research has demonstrated that emotion plays a key role in human decision making. Across a wide range of disciplines, old concepts, such as the classical ``rational actor" model, have fallen out of favor in place of more nuanced models (e.g., the frameworks of behavioral economics and emotional intelligence) that acknowledge the role of emotions in analyzing human actions. We now know that context, framing, and emotional and physiological state can all drastically influence decision making in humans. Emotions serve an essential, though often overlooked, role in our lives, thoughts, and decisions. However, it is not clear how and to what extent emotions should impact the design of artificial agents, such as social robots. In this paper I argue that enabling robots, especially those intended to interact with humans, to sense and model emotions will improve their performance across a wide variety of human-interaction applications. I outline two broad research topics (affective inference and learning from affect) towards which progress can be made to enable ``affect-aware" robots and give a few examples of applications in which robots with these capabilities may outperform their non-affective counterparts. By identifying these important problems, both necessary for fully affect-aware social robots, I hope to clarify terminology, assess the current research landscape, and provide goalposts for future research.

One of the primary realms into which artificial intelligence research has ventured is that of psychometric tests. It has been debated since Alan Turing proposed the Turing Test whether performance on tests should serve as the metric by which we should determine whether a machine is intelligent. This is an idea that may either solidify or challenge, depending on the reader's predisposition, one's sense of what artificial intelligence really is. As will be discussed in this paper, there is a history of efforts to create agents that perform well on tests in the spirit of an interpretation of artificial intelligence called ``Psychometric AI''. However, the focus of this paper is to describe a machine agent, hereafter called the CFE Agent, developed in this tradition. The CFE Exam is a gateway to certification in the Association of Certified Fraud Examiners (ACFE), a widely recognized professional credential within the fraud examiner profession. The CFE Agent attempts to emulate the successful performance of a human test taker, using what would appear to be simplistic natural language processing approaches to answer test questions. But it is also hoped that the the reader will be convinced that the same core technologies can be successfully applied within the larger domain of fraud detection. Further work will also be briefly discussed, in which we attempt to take these techniques to the next level, a deeper level, by which we can get a better sense of the knowledge the agent is using, and how that knowledge is being applied to formulate answers.

We present Saul, a new probabilistic programming language designed to address some of the shortcomings of programming languages that aim at advancing and simplifying the development of AI systems. Such languages need to interact with messy, naturally occurring data, to allow a programmer to specify what needs to be done at an appropriate level of abstraction rather than at the data level, to be developed on a solid theory that supports moving to and reasoning at this level of abstraction and, finally, to support flexible integration of these learning and inference models within an application program. Saul is an object-functional programming language written in Scala that facilitates these by (1) allowing a programmer to learn, name and manipulate named abstractions over relational data; (2) supporting seamless incorporation of trainable (probabilistic or discriminative) components into the program, and (3) providing a level of inference over trainable models to support composition and make decisions that respect domain and application constraints. Saul is developed over a declaratively defined relational data model, can use piecewise learned factor graphs with declaratively specified learning and inference objectives, and it supports inference over probabilistic models augmented with declarative knowledge-based constraints.We describe the key constructs of Saul and exemplify its use in developing applications that require relational feature engineering and structured output prediction.

Our work is aimed at service robots deployed in human environments that will need many specialized object manipulation skill. We believe robots should leverage end-users to quickly and efficiently learn the affordances of objects in their environment. Prior work has shown that this approach is promising because people naturally focus on showing salient rare aspects ofthe objects (Thomaz and Cakmak 2009). We replicate these prior results and build on them to create a semi-supervised combination of self and guided learning.We compare three conditions: (1) learning through self-exploration, (2) learning from demonstrations providedby 10 naive users, and (3) self-exploration seeded with the user demonstrations. Initial results suggests benefits of a mixed initiative approach.

Partially observable stochastic games (POSGs) are a robust and precise model for decentralized decision making under conditions of imperfect information, and extend popular Markov decision problem models. Complexity results for a wide range of such problems are known when agents work cooperatively to pursue common interests. When agents compete, things are less well understood. We show that under one understanding of rational competition, such problems are complete for the class NEXP^NP. This result holds for any such problem comprised of two competing teams of agents, where teams may be of any size whatsoever.

In this paper we present a technique for procedurally generating sections of 3D level geometry using computational evolution and guided by the visibility of certain game objects or areas during play. We show that certain level design goals can be achieved in the resulting levels, such as encouraging or dissuading player sightings of certain objects or locations. We also give details of a simple study of players on the generated levels, and discuss how this might be expanded to incorporate more complex problems related to level design.