AI training -- going beyond the cryptoverse

Was thinking about how AI is trained, and would love to know that the TokenMetrics AI is being trained on other factors beyond the fundamentals, tech, etc of crypto, as if it doesn’t exist in a vacuum. For example, factoring in international fiat currency prices, S&P, independent stocks, major world events (harder to quantify I know). It’s possible amidst all that data some other predictive factors might be found.

AI is only as good as what it’s trained on. The classic miss is google image search AI being trained on white people and then not being able to identify black faces. Or identifying black women inaccurately as men. So just wanted to ask the AI/engineering team about this.

1 Like

Hey Joe, currently our AI only does analysis based on crypto metrics. One exciting feature we’re working on adding though is a sentiment analysis. This metric will take data from Twitter and compile a score for a token based on how much engagement said token receives on the social media platform.