CAVAA

Working towards artificial awareness to improve human-machine interaction

The CAVAA project (Counterfactual Assessment and Valuation for Awareness Architecture) works to enhance the user experience in human-technology interactions by improving the knowledge of awareness in biological systems and engineering some of its components and dimensions into technological systems such as robots and artificial agents. Ultimately, CAVAA aims to enable artificial systems with a state of awareness that would allow them to understand their surrounding environment.

The CAVAA project is developing a specific computational framework, with an architecture that consists of key components of awareness, such as perception, memory, virtualization, simulation, and integration. Against this background, researchers in the project will then conduct experiments across various scenarios involving interactions between multiple humans and artificial agents. These scenarios include tasks like robot foraging, social robotics, and computer game benchmarks. The project aims to address trade-offs, such as balancing search efficiency and robustness, while also ensuring the acceptance of human users of this potentially "aware" technology.

Philosophy and ethics play an important role in CAVAA, with a dedicated effort considering shared ontologies, intention complementarity, behavioural matching, empathy, relevance of outcomes, reciprocity, counterfactuals, and projections toward future scenarios. Among other things, the philosophical reflection contributes to defining benchmarks for artificial awareness, and the ethical foundation extends to predicting the impact of choices made within the framework of trustworthy AI. This work is led by researchers at Uppsala University's Centre for Research Ethics & Bioethics.

CAVAA's expected outcome is an improved user experience marked by clarity, adaptability, and understanding. Ultimately, the project aims to make an impact on how we see the relationship between humans, other species, and smart technologies, and deepen our understanding of what the presence of awareness in artificial systems might mean, theoretically and practically.