Alexandr Wang at AI Summit: Personal AI Vision Matches NEAR's Philosophy

Captions as YouTube publishes them, usually machine-generated, so product and people names can be misheard.

1. 0:00

   Good morning Legion. We kick off this weekend. Video before NEARCON with Alexander Wang discussing the future of AI at the AI Impact Summit in India and sounding a lot like NEAR co-founder Illia Polosukhin You can listen in for yourself.

2. 0:14

   Our vision is personal superintelligence AI that knows you, your goals, your interests, and helps you with whatever you're focused on doing. It serves you, whoever you are, wherever you are. We all lead busy lives. I'm sure you'd want to do more, if only you at the time and headspace.

3. 0:37

   That's how I think about personal superintelligence. Say you want to be healthier. Your personal AI can help you see through a personal health plan covering diet, exercise and sleep and your daily routine. Or you have a project you'd like to get done, like putting on an event.

4. 0:55

   You can track your progress, reach out to venues, arrange invites, remind you of things you haven't considered, and more. If you love to go fishing or paint or want to travel more, you can help free you up so you can do more of these things and you can give you advice when you need it and or help you show up as a better friend or in your community.

5. 1:15

   It won't just do your admin, it'll be an extension of you so you can be you more. You see personal superintelligence that is yours and yours alone was Illia’s original vision for user owned AI and written about and spoken about many times before.

6. 1:32

   We at the Legion are actually in the process of finalizing a book on this exact topic, with transcripts of Illia’s as many talks that synthesize this vision for the first time. Be on the lookout for that after NEARCON. Next up, a nice news day for specialized models.

7. 1:48

   This 380 million parameter model called Zuna, recently dropped. It reconstructs missing brain signals from partial EEG data. It works across any electrode set up from consumer headsets to 256 channel research systems without retraining.

8. 2:04

   It lets you reconstruct missing EEG channels from sparse data, denoise corrupted signals, predict new channels from just electrode coordinates, and handle arbitrary electrode layouts. It's an incredible example of what specialized fine tuned models will be able to do in the future.

9. 2:20

   It's open source on hugging face, and as Velco Dar says, they're accelerating innovation and thought to text communication and making advanced neurotechnology more accessible to researchers worldwide. Expect much more of this in the year to come.

10. 2:35

    This is a future we can be very excited about. We'll do this all again real soon.