> ML was solving real problems from day one
No, machine learning developped slowly over decades before it could be fit into practical applications.
Bitcoin's launch was more than the launch of a product, it was the first time the ideas in its paper were actually put together. If we put cryptocurrency on the ML R&D timeline, we'd be in 1970 right now.
Of course, in the present day it solves more niche problems than ML which everybody uses through Google, but the fact that it's already used by real people at all is remarkable.
Note that the ML comparison was specific to the last decade mentioned by the person I replied to, not a history of the field. It’s possible that in the future we’ll look back at this period similarly to the first AI winter but I think even that would be disappointing to most participants because whatever succeeds won’t make them rich the way everyone in the space is hoping.
Just like with any nascent, hyped-up field, you will have people who are in it purely for the money. The fact that cryptocurrency makes it trivial to operate and participate in Ponzis exacerbates the issue, but if you went to a random ML/AI conference a few years ago you'd also have been pitched dozens of Ponzis (in the form of nonsensical startups) by opportunists who understand nothing about the technology.
[1]: https://en.wikipedia.org/wiki/Craig_Steven_Wright