I have to say I'd be a bit stumped to predict what the CAD system of 2020 or 2030 will be like

I've been writing about CAD software since the days when everything came on a five-and-a-quarter-inch floppy. I remember a press conference in a pub introducing AutoCAD Release 9, the successor to 2.6, at a point when that product had begun to dominate the booming PC-based CAD market. And if you were there, you'd never forget the chaos which followed UK developer PAFEC's decision to respond by giving its own PC-DOGS product away for free.
This was an age where 30Mb hard drives had just begun to appear on PCs, but the DOS operating system couldn't cope with more than 20Mb, so you had to format the drive into two partitions. Nice. CAD was not only the most interesting subject for us engineering journalists to write about, it was everything. The CAD companies took over the UK's design engineering exhibition (DES), pushing the component manufacturers to the periphery, and when the software companies wound down their marketing in the 90s, the show never recovered.
However primitive those early programs were, however, I still doubt that today's solid modelling applications would have surprised us back then. Stunning as some of them are, I suspect they're exactly what we'd have expected to see, 15 to 20 years on. I'm sure there are some old articles I can dig out where some of us had a go at predicting what CAD systems in 2000 or 2010 would be like, and I'm sure they won't be far off the mark - except I'm sure they'd have predicted a stereolithography machine on every desktop by now. Sadly, whilst the technology's here, the market isn't.
But I have to say I'd be a bit stumped to predict what the CAD system of 2020 or 2030 will be like. Where else is there to go? 3D holograms, I guess. Intelligent virtual designers which require just some parameters to come up with optimal or alternative designs? I'll have to organise a pub conference with some readers and discuss it sometime.