1,581 karma · joined February 2, 2012
The trick there is that they both have related physical analogs, and machine learning math doesn't really (in that while you can visualize them spatially, it doesn't seem to help solve any problems in that space.)
(Of course, the place I've dealt with this the most was a former post office building from the late 1930s - we didn't really need tests, a historian would probably have just said "of course you have asbestos" :-) It does expose an interesting problem with the labeling part, though: USPS as an ancient federal agency has its own rules for things and had a distinct asbestos markers - an A in a point-down triangle - which we fortunately found documentation for, it isn't public record.)
(I'm still happy to have traded that for planes with a much higher cabin pressure-altitude - and would trade even further for business-class transatlantic non-cruise ships. But it was definitely fun to have had the experience.)
(also: hasn't the x4-v2 already been announced that is supposed to have a frontlight?)
I think that might be just imagination - android autocorrect in particular got sufficiently worse that I finally turned it off (I still use it as a "typing assist" - it only displays choices that I can tap to replace, or (more often) ignore.)