🧠 Food for thought

Implications, context, and why it matters.

The paper uses a small dataset, even as Meta mentions 700 plus volunteers for TRIBE v2

• Meta says TRIBE v2 uses functional magnetic resonance imaging (fMRI) recordings from more than 700 healthy volunteers who viewed images, listened to podcasts, watched videos, and read text, yet the scientific paper in the source material covers a narrower set of experiments 1.
• Those experiments rely on more than 80 hours of fMRI data per person from six participants watching the TV show Friends plus four movies 1.
• The TRIBE model won the Algonauts 2025 competition (a research challenge focused on predicting brain responses from stimuli) after it predicted brain activity for this group with high accuracy, including on out-of-distribution material such as silent black-and-white movies 1.
• That training mix stays limited and specific, which raises questions about how well the results carry over to broader populations and everyday settings 1.

The “digital twin” idea is drawing excitement and worry

• Meta and others describe TRIBE v2 as a generalizable “digital twin” of human brain responses, which could support computational experiments that speed up research 2.
• Online discussion has drifted toward possible commercial uses including drug discovery and neurotechnology, plus one commenter said they plan to send a formal LOI/LOA/MOU package (letter of intent/letter of authorization/memorandum of understanding) to Meta to explore collaboration 2.
• Some reactions call the technology “terrifying,” which speaks to anxiety about AI that can predict brain activity 2.
• Meta’s Creative Commons BY-NC license allows reuse with attribution for noncommercial purposes, which may slow near-term commercial rollout, yet it does not direct the work toward any single goal such as AGI [NEWS ARTICLE].