My first degree was in a foreign language; I was taking it for my own reasons but one of the professors I was close to knew I was very involved in technology and linguistics. She asked in, maybe in 1992, whether I thought computers would eliminate the jobs of translators "soon."

I remember looking at her and saying that no, not until the mid-2000s or so, because of the twin problems of voice recognition and OCR: in both cases, you need both a corpus for translation and to build it you need to get the data into machine form.

I was off by maybe 6-10 years but machine translation really did start making inroads for professional writing around then (and maybe a little sooner). Now, though, the job is totally transformed. Live translation is still a real job but a hard one to land and with fewer opportunities, and media translation while it is still a career (especially for science and technology papers, etc.) it is dramatically changed and is now more of an editing job than a translation job. That happened in a few years, tops, once the tools arrived.

So .. maybe we're whistling in the dark. A lot of what we do cannot really ever be practically done by AI, but I think a lot could be. The demise of the CASE market probably delayed a lot of what's coming for a long time, and a lot of things likely will be automated. I think UI design is ripe since so many UIs basically look exactly the same.