# Generative BCI Decoding **Domain:** Brain-Computer Interfaces / Generative AI **Doc Type:** Emerging Concept Node **Maturity:** Research Frontier **Related:** [[Neural Decoding]], [[Cross-Modal Neural Decoding]], [[IEEE P3766]], [[Multimodal Neural Data Fusion]] --- ## Definition **Generative BCI decoding** uses generative models to reconstruct or produce semantically rich outputs—such as text, speech, images, or other structured representations—from neural signals. The model does not simply select a class; it generates an output conditioned on decoded neural evidence and learned priors. ## Risk boundary Generative priors can improve legibility while also introducing plausible content not supported by the recorded signal. Provenance, confidence, calibration, and the separation of measured evidence from model completion are therefore essential.