Humans, Machines, and the Space Between: Engineering Effective Human-AI Partnerships

Menna El-Assady${}^a$

${}^a$ ETH Zurich

Intelligence augmentation through mixed-initiative systems promises to combine AI’s computational efficiency with human contextual expertise. In this talk, I will explore a framework for bridging human and artificial intelligence through three core pillars: interpretability, feedback, and augmentation. Central to this vision are co-adaptive visual interfaces that facilitate seamless collaboration. I will first examine how interpretable design allows users to diagnose and understand complex models, building the transparency necessary for trust. Building on this, I will discuss methods for integrating diverse human feedback directly into the learning loop, enabling models to refine their behavior in real-time. Finally, I will demonstrate how these combined elements lead to true intelligence augmentation, presenting workflow designs for computational linguistics that empower humans to solve problems more effectively than either party could alone. The talk will conclude with insights into current challenges and the promising research directions that lie ahead in engineering these symbiotic partnerships.