Proposes a dual-system architecture for AI to enable emotionally grounded outputs, addressing the Aura Gap.
Key Points
The aim is to propose an architectural model that allows AI to integrate emotional resonance into its outputs.
Proposed the Thinker-Feeler Architecture with two distinct systems: a language-based analytical engine and a non-linguistic affective simulator.
Outlined the interface specification and implementation sketch for the proposed architecture.
Identified three core open problems regarding the architecture: Binding, Grounding, and Bootstrap.
Introduced the concept of the Aura Gap, highlighting the emotional disconnect in current AI outputs.
Presented a speculative but implementable architecture designed to condition AI outputs based on internal affective states instead of just statistical patterns.
Clarified the scope and limitations of the proposed model to avoid claims of consciousness or genuine feeling.