I’ve taken a look at a couple of articles I’ve found, and really I still fail to miss the potentially big thing about combining the two, more than the current stand of AI processing information on large centralized or decentralised databases.
From the AIs perspective, Blockchain can be considered in essence as a database repository (non- tamperable, distributed, and all we like), so AI can be applied to whatever information resides there, just as with any other DB repository. The context of application is derived from the nature of the information stored in the DBs or on the specific Blockchain. It looks like (evolving) business as usual, but including user case scenarios where the information resides on given Blockchain instead on more regular DBs. There are obvious applications, but due to where the data resides.
Some related articles which fail to provide, in my opinion, a reason to see this as a global breakthrough:
https://emerj.com/ai-sector-overviews/ai-in-blockchain/https://thenextweb.com/hardfork/2019/02/05/blockchain-and-ai-could-be-a-perfect-match-heres-why/This does sound fantastical but the one possible use of blockchain could be the way ML models get their data. I read something of this sort just today on Lightning Dev Alex Bosworth's twitter.
https://twitter.com/alexbosworth/status/1160936346833174528He talks about training Human fascimile models. The concept of Human fascimile model looks similar to the "Digital Twin" model for machinery and equipment. You need large amount of data regarding choices humans make in order to train your model. The more data you have, better the model would be.
Though he only talks about paying people using Lightning for this job, yet, i feel the tweet also points towards using the obvious database of bitcoin transactions that consists of a humungous number of choices of what is being spent where.
The tweet is pretty cryptic and hard to understand. No doubt it received such less attention. Yet, Its an interesting line of thought the way i could interpret.