Some are

Researchers have previously decoded brain activity representing words or images before, and even a word or two for videos.

But this study takes a step in getting more accurate longer descriptions based on mental representations of video, which are quite complex!

www.science.org/doi/10.1126/...

You can see more results below.

LEFT: The videos shown to participants are on the left.
RIGHT: the AI's descriptions after 100 iterations interpreting brain activity from each of the 6 participants.

This is part of ongoing efforts to improve interpretation tools for people with language difficulties like aphasias.

The researchers had 6 different people watch or recall videos while getting an fMRI. Then they tested whether a suite of AI tools could translate the fMRI data into descriptive text.

Video captured by microscopes Jay McClellan, who took first prize in the 2025 Nikon Small World in Motion Competition. Shared with permission & provided by Nikon Small World.

Watch the full videos and see all the stunning honourable mentions: www.nikonsmallworld.com/galleries/20...

Every year!!

PS. if you want more details, I broke it down in a new video: www.instagram.com/p/DOYkStDjd2g/ (with receipts!)

You all were the BEST audience!!! Our team is so happy so many public servants showed up and showed out!!!

Thanks for listening!! Really appreciate it 🙏🏼

Hi Victor! Glad you were interested in the episode -- it's a segment in the episode titled, "The Scramble to Solve Space Scrap"

It's mentioned in the episode description but pretty subtly so very easy to miss -- I should've pointed the link to the exact episode so it's easier to find, apologies!