Using a cognitive architecture to consider antiBlackness in design and development of AI systems

TL;DR

This paper explores how cognitive modeling, specifically using ACT-R and ConceptNet, can reveal the influence of antiblackness and racism in AI system design, highlighting the sociocultural biases embedded in cognitive architectures.

Contribution

It introduces a novel approach to incorporate sociocultural and racial biases into cognitive modeling of AI systems using ACT-R and ConceptNet.

Findings

Cognitive architectures often overlook sociocultural biases.

Modeling antiblackness reveals hidden biases in AI development.

Connecting sociocultural context with cognitive models can improve AI fairness.

Abstract

How might we use cognitive modeling to consider the ways in which antiblackness, and racism more broadly, impact the design and development of AI systems? We provide a discussion and an example towards an answer to this question. We use the ACT-R/{\Phi} cognitive architecture and an existing knowledge graph system, ConceptNet, to consider this question not only from a cognitive and sociocultural perspective, but also from a physiological perspective. In addition to using a cognitive modeling as a means to explore how antiblackness may manifest in the design and development of AI systems (particularly from a software engineering perspective), we also introduce connections between antiblackness, the Human, and computational cognitive modeling. We argue that the typical eschewing of sociocultural processes and knowledge structures in cognitive architectures and cognitive modeling implicitly furthers a colorblind approach to cognitive modeling and hides sociocultural context that is always present in human behavior and affects cognitive processes.