[Backlog | Draft] And so it begins ... AGI + Metaverse!

On Jan. 18, Zuck took it to Insta to announce the merge of two major AI research efforts at Meta (FAIR and GenAI), in an effort to achieve the long term goal, which he said is

"to build general intelligence, open source it responsibly, and make it widely available."

Of course, there wasn't enough detail on what the open source release policy will look like, and how they will determine when it's responsible.

He provided the following specifics:

1. By the end of the year, they will have 350,000 NVIDIA H100s or around 600,000 H100 equivalents of compute. Some have suggested that this could be 6 times more powerful than the current supercomputer frontrunners. That's CRAZY! But yes, Zuck sees it.

2. They currently training llama 3 and we can expect new models.

3. They understand the significance for AI devices and will be working on glasses for AI-human communication. This is very interesting (think Tony Stark glasses). It's also very exciting. The AI will receive ALL the vision and audio input we are able to receive! Lots of applications and use-cases. Advertisers will be jumping onto this. But more importantly, A LOT MORE real world data, which can make these AGI models much more powerful! They've collaborated with RayBan for this, which seems like the best possible partnership.

Overtime, the lines between Oculus and Smart glasses will blur. I think the leap to Glass + AGI drastically scales up the NEED for Neuralink to gain more resources and deliver. It will be interesting to see Neuralink's roadmap for the next 5 years.

One thing is for sure, Zuck going all in on AGI is great because we have more resources, players, and competition in the game. An advancement in any team will be leveraged by other teams, and it's a virtuous cycle that will lead to exponential advancements in all of tech. However, I would love to see emphasis on what the plan is to open-source.

It's crazy that these huge tech companies are pouring all their resources into creating AGI and yet the public has yet to see the details of how these systems work specifically. Security should consider all use-cases. People have very complicated definitions of security. How do these teams plan to achieve it? Shouldn't they collaborate at least on the security necessities and requirements and make it as rigorous as possible?