To engage in human-like dialogue, robots require the ability to describe the objects, locations, and people in their environment, a capability known as "Referring Expression Generation." As speakers repeatedly refer to similar objects, they tend to re-use properties from previous descriptions, in part to help the listener, and in part due to cognitive availability of those properties in working memory (WM). Because different theories of working memory "forgetting" necessarily lead to differences in cognitive availability, we hypothesize that they will similarly result in generation of different referring expressions. To design effective intelligent agents, it is thus necessary to determine how different models of forgetting may be differentially effective at producing natural human-like referring expressions. In this work, we computationalize two candidate models of working memory forgetting within a robot cognitive architecture, and demonstrate how they lead to cognitive availability-based differences in generated referring expressions.

Inspired by e-participation systems, in this paper we propose a new model to represent human debates and methods to obtain collective conclusions from them. This model overcomes drawbacks of existing approaches by allowing users to introduce new pieces of information into the discussion, to relate them to existing pieces, and also to express their opinion on the pieces proposed by other users. In addition, our model does not assume that users' opinions are rational in order to extract information from it, an assumption that significantly limits current approaches. Instead, we define a weaker notion of rationality that characterises coherent opinions, and we consider different scenarios based on the coherence of individual opinions and the level of consensus that users have on the debate structure. Considering these two factors, we analyse the outcomes of different opinion aggregation functions that compute a collective decision based on the individual opinions and the debate structure. In particular, we demonstrate that aggregated opinions can be coherent even if there is a lack of consensus and individual opinions are not coherent. We conclude our analysis with a computational evaluation demonstrating that collective opinions can be computed efficiently for real-sized debates.

An ambitious goal for artificial intelligence is to create agents that behave ethically: The capacity to abide by human moral norms would greatly expand the context in which autonomous agents could be practically and safely deployed. While ethical agents could be trained through reinforcement, by rewarding correct behavior under a specific moral theory (e.g. utilitarianism), there remains widespread disagreement (both societally and among moral philosophers) about the nature of morality and what ethical theory (if any) is objectively correct. Acknowledging such disagreement, recent work in moral philosophy proposes that ethical behavior requires acting under moral uncertainty, i.e. to take into account when acting that one's credence is split across several plausible ethical theories. Inspired by such work, this paper proposes a formalism that translates such insights to the field of reinforcement learning. Demonstrating the formalism's potential, we then train agents in simple environments to act under moral uncertainty, highlighting how such uncertainty can help curb extreme behavior from commitment to single theories. The overall aim is to draw productive connections from the fields of moral philosophy and machine ethics to that of machine learning, to inspire further research by highlighting a spectrum of machine learning research questions relevant to training ethically capable reinforcement learning agents.

The ability to cooperate through language is a defining feature of humans. As the perceptual, motory and planning capabilities of deep artificial networks increase, researchers are studying whether they also can develop a shared language to interact. From a scientific perspective, understanding the conditions under which language evolves in communities of deep agents and its emergent features can shed light on human language evolution. From an applied perspective, endowing deep networks with the ability to solve problems interactively by communicating with each other and with us should make them more flexible and useful in everyday life.

Embodied AI has been recently gaining attention as it aims to foster the development of autonomous and intelligent agents. In this paper, we devise a novel embodied setting in which an agent needs to explore a previously unknown environment while recounting what it sees during the path. In this context, the agent needs to navigate the environment driven by an exploration goal, select proper moments for description, and output natural language descriptions of relevant objects and scenes. Our model integrates a novel self-supervised exploration module with penalty, and a fully-attentive captioning model for explanation. Also, we investigate different policies for selecting proper moments for explanation, driven by information coming from both the environment and the navigation. Experiments are conducted on photorealistic environments from the Matterport3D dataset and investigate the navigation and explanation capabilities of the agent as well as the role of their interactions.

Power Virtual Agents is gaining traction around the world and the market has responded with a strong desire for us to support more languages. Today we're excited to announce that we are bringing to public preview support for an extended set of languages! This enables our partners and customers to build even more engaging and locally relevant experiences for their users. When you create a new bot, you select the language you want the bot to understand when interacting with your users. You'll see that your new bot is prepopulated with content in the target language and you can easily create more topics with trigger phrases and bot responses in the language you've selected.

Reuse is a common system architecture approach that seeks to instantiate a system architecture with existing components. However, reusing components with AI capabilities might introduce new risks as there is currently no framework that guides the selection of necessary information to assess their portability to operate in a system different than the one for which the component was originally purposed. We know from SW-intensive systems that AI algorithms are generally fragile and behave unexpectedly to changes in context and boundary conditions. The question we address in this paper is, what type of information should be captured in the Interface Control Document (ICD) of an AI-enabled system or component to assess its compatibility with a system for which it was not designed originally. We present ongoing work on establishing an interface description template that captures the main information of an AI-enabled component to facilitate its adequate reuse across different systems and operational contexts. Our work is inspired by Google's Model Card concept, which was developed with the same goal but focused on the reusability of AI algorithms. We extend that concept to address system-level autonomy capabilities of AI-enabled cyber-physical systems.

The aim of this article is to delineate a theory of interaction semantics and thereby provide a proper understanding of the "meaning" of the exchanged characters within an interaction. The idea is to describe the interaction (between discrete systems) by a mechanism that depends on information exchange, that is, on the identical naming of the "exchanged" characters -- by a protocol. Complementing a nondeterministic protocol with decisions to a game in its interactive form (GIF) makes it interpretable in the sense of an execution. The consistency of such a protocol depends on the particular choice of its sets of characters. Thus, assigning a protocol its sets of charaacters makes it consistent or not, creating a fulfillment relation. The interpretation of the characters during GIF execution results in their meaning. The proposed theory of interaction semantics is consistent with the model of information transport and processing, it has a clear relation to models of formal semantics, it accounts for the fact that the meaning of a character is invariant against renaming and locates the concept of meaning in the technical description of interactions. It defines when two different characters have the same meaning and what an "interpretation" and what an "interpretation context" is as well as under which conditions meaning is compositional.

We propose a multi-agent variant of the classical multi-armed bandit problem, in which there are N agents and K arms, and pulling an arm generates a (possibly different) stochastic reward to each agent. Unlike the classical multi-armed bandit problem, the goal is not to learn the "best arm", as each agent may perceive a different arm as best for her. Instead, we seek to learn a fair distribution over arms. Drawing on a long line of research in economics and computer science, we use the Nash social welfare as our notion of fairness. We design multi-agent variants of three classic multi-armed bandit algorithms, and show that they achieve sublinear regret, now measured in terms of the Nash social welfare.

We consider the problem of finding patrol schedules for $k$ robots to visit a given set of $n$ sites in a metric space. Each robot has the same maximum speed and the goal is to minimize the weighted maximum latency of any site, where the latency of a site is defined as the maximum time duration between consecutive visits of that site. The problem is NP-hard, as it has the traveling salesman problem as a special case (when $k=1$ and all sites have the same weight). We present a polynomial-time algorithm with an approximation factor of $O(k^2 \log \frac{w_{\max}}{w_{\min}})$ to the optimal solution, where $w_{\max}$ and $w_{\min}$ are the maximum and minimum weight of the sites respectively. Further, we consider the special case where the sites are in 1D. When all sites have the same weight, we present a polynomial-time algorithm to solve the problem exactly. If the sites may have different weights, we present a $12$-approximate solution, which runs in polynomial time when the number of robots, $k$, is a constant.