Thread by @swardley

X : Will AI do research?
Me : Odd question. AI is already involved in research - from literature searches to finding correlations to even creating hypothesis. It's quite far from the general purpose intelligence you need for causation but maybe in fifty years or so. Why?

X : How do you industrialise research?
Me : Well, you can industrialise components around research but if you're talking about genuine core research then it's in that genesis phase. You literally can't industrialise it.
X : What bits can you industrialise?

Me : Any surrounding components that use defined / known models i.e. publishing, finance, marketing, legal (contracts to patents), HR, admin ... even things like sales. Those would seem like obvious targets. You'd have to map it.

X : Patents?
Me : Yep. That's in the area of applied research as opposed to core research i.e applied takes the concepts from core and applies it to some thing. You could probably make good mileage out of AI to do applied research and create patents.

Me : As a general rule, Gov funds core research which progresses society. Corporations fund applied research which exploits core research for financial gain. You could more easily apply AI to applied but without core then as the source dries up, so does the downstream.