I know it sounds cool but it doesn't work most of the times. These tricks are useless as 98% of the stuff Reddit has produced.
You need some tool for ideation more than finding some shit in datasets.
I think if you want really see good patterns that you can utilize and pump your wallets, my idea would be to create social media website simulation and then find patterns / vulnerabilities in that simulation. That's probably a mix of some supervised / unsupervised learning, but I'm not an expert.
Would be interesting to see patterns in data nobody else can see, huh.
Well, I think to make it more feasible it can be just scraping of YouTube / using its data API and specific niche. Observe some patterns over some time under select keywords using machine learning and yeah, could be some insight, but I'm not really sure!
I mean, if you program it yourself, you are almost certain that it's built for your use case and that you can trust it cause you know how good or bad that something you created is.
---
What if you created AI that actually understood you and could fix your errors before they happened? $666.666.666 awaiting.
Something smarter than you that could make you make 500% better decisions. That sh1t would be sick.