AIR-O decodes visual attention from brain scans in real time, generating images noninvasively

jonxuxu · x · 2026-11-18

Alljoined unveils AIR-O, a model that decodes what a person is attending to directly from noninvasive brain scans, generating an image from scratch in real time that depicts where their attention is focused.

Author jonxuxu recalls that when the first fMRI-to-image models appeared in 2022, his questions were "is this real?" and "how far can we take this?"—three years of work later, the team is sharing its results.

He argues BCI will play a role in helping exponentially more powerful AI align with humanity's best interests.

Original post →

More from Research

Research channel →