The Perception Research and Artificial Development Conference

The Perception Research and Artificial Intelligence Development (PRAID) Conference aims to spearhead a collaborative effort to unravel the complexities of AI perception and align it more closely with human understanding. PRAID 2024 was the inaugural PRAID conference, hosted at Queen’s University in April.  The conference featured engaging series of talks, panels, and discussions with experts in Computer Science, Neuroscience, and Philosophy. This event served as a critical platform for collaboration and innovation, addressing the intricate challenges surrounding AI perception.

             The conference featured an interdisciplinary and diverse lineup of speakers from universities across Canada and globally. Attendees had the unique opportunity to engage with experts from computing, philosophy, psychology, neuroscience, and more. Our speakers included esteemed local scholars from Queen’s University, as well as international thought leaders you may be discovering for the first time here.

The Perception Problem

At the heart of modern artificial intelligence (AI) development lies a critical challenge known as “The Perception Problem.” This issue addresses the fundamental gap between the capabilities of AI systems to gather information and their ability to discern what is truly relevant in a dynamic environment. For instance, an autonomous vehicle must be able to differentiate a child crossing the street from irrelevant background elements like parked cars or distant buildings. 

           Despite significant advancements in AI, achieving this level of discernment remains a daunting task, largely due to the inherent complexities of mimicking human-like perception. The Perception Problem also raises important questions about safety standards in AI development, especially in contexts where making accurate and consistent perceptual judgments is crucial.

           The current AI systems that are primarily based on deep learning techniques, have shown remarkable success in various domains. However, they often exhibit a “black-box” behaviour, complicating their evaluation in safety-critical applications. Furthermore, these systems are susceptible to adversarial errors, where minor alterations in input can lead to drastically incorrect outputs, highlighting a fundamental disconnect between AI and human perception.

           Addressing The Perception Problem requires a concerted effort to integrate insights from diverse fields including neuroscience, psychology, and philosophy. It calls for a reevaluation of the underlying models of perception that guide AI development, moving beyond traditional representational views to consider more dynamic and interactive frameworks like the sensorimotor approach. 

Ongoing Collaborative Efforts and the Future

The collaborative spirit of the conference continues, as the researchers who spoke and participated in PRAID are now writing a comprehensive white paper on AI perception. This white paper will encapsulate the collective expertise and innovative ideas discussed at the conference, providing a valuable resource for future research and development in the field.

The PRAID Conference 2024 at Queen’s University was a milestone event in the ongoing quest to align AI systems more closely with the nuanced mechanisms of human perception. The interdisciplinary collaboration and innovative discussions held at the conference promise to drive future advancements in AI perception, making strides towards creating AI that truly understands and interacts with the world as humans do. The forthcoming white paper will further solidify the insights gained and chart the path for future breakthroughs.

In the future, we hope to build on the success of the 2024 PRAID Conference by hosting more events dedicated to exploring AI perception. Our aim is to continue fostering an environment of interdisciplinary collaboration and innovation, bringing together experts from various fields to share their insights and advance our understanding of AI perception. These future conferences will provide ongoing opportunities for knowledge exchange, ensuring that we stay at the forefront of this rapidly evolving field.

Scroll to Top