#vass — Public Fediverse posts
Live and recent posts from across the Fediverse tagged #vass, aggregated by home.social.
-
To ChatGPT: If Android XR smartglasses become affordable mass market products, how will this affect adoption of hybrid use of AI scene description and visual-to-auditory sensory substitution by blind people? https://chatgpt.com/share/6a630864-ac58-83ed-9e33-dee8380b0f8d #AI #VASS #BCI #blindness
"The AI could become an always-available instructor: “The high-pitched vertical pattern is the left edge of the doorway. Turn your head slightly right.”" #blind #a11y
ChatGPT's answer archived at https://www.artificialvision.com/archive/chatgpt_SSD_24july2026.pdf (PDF file)
-
Perceptions of visual-to-auditory sensory substitution (VASS) technology among students with visual impairments and special educators https://journals.sagepub.com/doi/10.1177/02646196261450976 "applies The vOICe algorithm to convert visual images into auditory input"; #ETRI #V2A #VASS #BCI #blindness
-
Paradromics blog post on intracortical BCIs vs. ECoG (electrocorticography) https://www.paradromics.com/blog/intracortical-versus-ecog-bci-modalities Note that for restoring vision, intracortical BCIs do not necessarily offer better performance than noninvasive VASS https://www.artificialvision.com/neuralink.htm #ICMS #VASS #BCI #NeuroTech
-
Research of this nature may prove tangential to the study of cross-modal plasticity in visual-to-auditory sensory substitution (#VASS) for the blind, while vice versa VASS can inform neuroscience of macroscopic brain changes from sensory deprivation conditions other than blindness or deafness.
It is not mainstream thinking yet, but VASS has the potential to be a noninvasive, affordable, sustainable and highly scalable alternative for brain implants for restoring vision to the blind - and it also works for congenitally blind. https://www.artificialvision.com/neuralink.htm Neuralink Blindsight anyone?
-
<ChatGPT>
Error modes: deterministic distortion vs hallucination#VASS errors
- Deterministic distortions (lighting, blur, occlusion) are consistent.
- User can learn compensations because the mapping doesn’t “make things up.”#LLM live vision errors
- Hallucination / confabulation risk: plausible but wrong descriptions.
- Overconfidence tone unless the UI forces uncertainty.
- Identity drift across turns: “a mug” becomes “a bowl” becomes “a jar.”
</ChatGPT>