This is a light-hearted look at a serious problem (see especially footnote [1]). What do academics working in artificial intellignce when corporate research throws millions of dollars into large scale infrastructure capable of training AIs with billions of data points? There's nothing a university can do to compete directly. The authors look at a list of alternatives, some serious, others less so, that academics can do to adapt. Personally, I think the only way smaller players can remain relevant is to do the one thing the authors don't suggest: decentralize. Work openly with others. It seems to be axiomatic that AI requires large centralized data centres running proprietary algorithms and data sets, but I don't see why that would be the case. And when the kind of supercomputing power that costs millions today is in the hands of the average person tomorrow, we're going to need to know how to interoperate.
Today: 7 Total: 102 [Direct link] [Share]
