In this research area, CITEC researchers explore how humans and A.I.-based systems learn and advance their abilities through interaction with their environment and with other agents. To achieve this in open environments requires the exploitation and integration of multiple learning strategies: learning (pro-)actively through self-driven exploration of the environment, continuously and in an online fashion through incrementally gained insight, and in interaction with knowledgable others that scaffold one's own learning. In CITEC, researchers from A.I./Machine Learning, Psychology/Pedagogy and Linguistics jointly carry out projects to investigate the principles and mechanisms of human learning and to develop intelligent systems that are capable of learning and tutoring in interaction.
Human cognition and communication are deeply rooted in our embodied multisensory and multimodal experience of the world. CITEC researchers study how we perceive and make sense of a complex open world through multiple modalities and how situated communication attains its effectiveness and efficiency through employing different modalities in remarkably versatile and parsimonious ways. CITEC research in this area spans all levels of linguistic analysis of polysemiotic communication by verbal and nonverbal means, as well as the processing, production and utilization of multimodal behavior and communication in A.I. systems.
Human-aware cognitive systems need to be able to navigate and live up to the social dimensions of our human nature. CITEC researchers explore how artificial agents can be made aware of the social aspects and structure of the environment they are placed in. This encompasses the sensitivity to and awareness of other agent's cognitive, intentional, dispositional or affective states, their relationships with other individuals and within groups, or the social expectations and norms that hold in specific contexts. The CITEC view also takes into account that, in interaction, social minds reciprocally model each other and that the resulting attributions and projections bear important consequence for the affordances but also ethical implications of interactions between humans and A.I. or robots.
Current A.I. system development often targets the introduction phase, aiming for high performance in a static and well-defined environment. Yet, a human-compatible system needs to preserve its functionality and maximize acceptance at the individual and societal level throughout its whole life-cycle, respecting short-term and long-term objectives, taking care of technological requirements and cognitive or societal impact, and doing so with reasonable maintenance costs, energy consumption, or load for humans interacting with it. CITEC researchers explore how A.I. systems can be made trustworthy and sustainable during their whole life cycle - addressing challenges such as transparency, explainability and fairness, as well as resource-efficiency of intelligent interactive systems during construction, application, maintenance and life-long learning.
Collaborative robots, autonomous vehicles, or smart environments - intelligent systems frequently figure in embodied forms to reside in physical environments shaped for and by humans. CITEC researchers study the psychological, cognitive, and engineering principles of how humans or artificial systems can interact purposefully and cooperatively with objects or agents in physical environments that are dynamic, complex and fuzzy. CITEC researchers study how human and biological agents learn to perceive and cognize about situations, and to structure and produce sensorimotor and cognitive behavior. Using this knowledge, CITEC develops robot systems or technologically augmented environments that interact with human users in cooperative ways and enable a new level of human-machine behavior.