Benefit From a Balanced Use of AI and Human Cognition
Human-Centered Data Analytics
Identifying patterns in Big Data is one thing; making sense of those patterns is another. While computers excel at the former, they lag significantly behind their human counterparts in the much harder problem of sensemaking. Kairos researchers are pioneering new methods for improving human analysts' ability to discover and make sense of the deeper causal structure lurking within noisy, uncertain data. We believe that by characterizing and improving human sensemaking, we will be in a better position to teach sensemaking skills to a new generation of assistive AI technologies.
Explainable Artificial Intelligence
Modern AI algorithms such as deep neural networks are often viewed as black boxes: although their performance is impressive, their inner workings and decision rationales are difficult to understand. This lack of transparency can lead to mistrust (or over-trust) of AI systems, hampering the ability of humans and machines to collaborate effectively. To meet this challenge, Kairos is teaming with researchers in academia to create new tools that automatically generate human-understandable explanations of an AI's "thought processes.” Armed with such explanations, a human operator can make informed judgments about whether to accept, modify, or overrule an AI’s decisions.
Adaptive Human-Machine Teaming
Increasingly, humans and machines work together to solve complex problems – whether in small tactical military teams or large-scale businesses and intelligence organizations. At Kairos we are developing new methods for instrumenting, assessing, and augmenting the coordinated problem-solving abilities of human and machine teams. Our approach blends insights from the fields of psychology, artificial intelligence, network science, and computational social science to optimize human-machine team performance in dynamic environments.