we are so early

I don't know anything about @ciniz but when I saw this white paper claiming that Onchain Gaias was going to democratize the develop AI, I had to call BS.

This is snake oil, written by someone who knows NOTHING about artificial intelligence, machine learning, deep learning, data science or anything else in that field.

I'm skeptical of what @ciniz is saying but this is a bold claim to make before anything has even launched.

Decentralized AI agents have been around for years (Fetch), and gamified incentive models work.

Acting like you have more of an idea than anyone else on what a valid AGI training model might like is wild.

Check this out, it will help rationalize:

father | artist | dev | crypto class of 2014 | DeFi enthusiast

2. They make a comment about compute being incredibly expensive, yet make no mention of how they plan to address that issue themselves. 

They literally do not know what they are talking about, it's snake oil, it's saying AI for the sake of AI bc people FOMO.

Still, without knowing what kind of final platform, game—or more likely from the context, many games—we're talking about, how can you say it will definitely not be useful as a training tool?

All of the bullet points in your breakdown could be presented to a ML system in a gamified manner.

Check this out, it will help rationalize:

https://paragraph.xyz/@mercerai/artificial-intelligence-in-4-minutes

Because AGI = Artificial General Intelligence, which means an AI that can do EVERYTHING a human can do, at an equal level, or better.

So how do you anticipate that a simple game, and pulling data from that to be used as "training" data, translates into doing everything, EVERYTHING a human can do?

What I'm saying is even the best ML engineers can't say what an AGI model looks like.

Without knowing what kind of games are involved, it's impossible to say if it's a valid training method. e.g. a social MMO is a totally different beast compared to chess.

I'm a machine learning engineer, lol.

The most basic premise is that there will be some game, which players take part of, will somehow create enough data to train / create artificial general intelligence. That's asinine. It's not how training a model works.