AI PCs
When all the tulips have withered and their bulbs have rotted in the ground AI augmented PCs might be repurposed for better optical character recognition systems (and handwriting recognition.)
Many decades ago I imagined systems that could be presented with a manuscript of technical writing which included equations, figures and text which the system could classify each textual component to produce an electronic version where the natural language text is represented as such, similarly equations, and figures as (vector?) graphics.
Basically the logical content, which could be processed to produce any number of presentation formats as well as being searchable or amenable to semantic processing.
(The semantic web was once(?) a thing but I have no idea what it was (or is?) as at the time I was too busying bofh~ing - every open window an unmissable opportunity. ;)
Not much of this has come to pass. I can scan a printed page from a journal but any software purporting to usefully process it is pretty piss poor at best.
Handwriting recognition isn't too bad when entering plain English text but try and enter anything technical like (∀x)... no dice. You are stuck with typewriter conventions like (x) or (Ax) and [x] or (Ex) which you might postprocess later or using Latex or eqn forms.
David Gries text The Science of Programming (1981) used (A x: ...) and (E x: ...) which definitely lacked a certain aesthetic elegance.
Not really the good old days but just that much has not improved, much has gone backwards and more has gone, as Cory Doctorow's coinage would have, to shit. So little time, so much sky to shout at (2π sr actually.:)
Biting the hand that feeds IT
