4 ms·

I was advised in 1989 that I shouldn't study CS because it'd be outsourced/replaced by AI. What did happen were waves of outsourcing, followed by insourcing. M

I was advised in 1989 that I shouldn't study CS because it'd be outsourced/replaced by AI. What did happen were waves of outsourcing, followed by insourcing. My theory is that whenever there's a new generation of managers, we flip around, because they haven't learned from the previous one.

foobiekr 7y ago
A somewhat related story. 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.
- groby_b 7y ago
  UI design will live for a long, loooong time, because every single customer is basically like "I'm somewhat of an expert myself". Preventing them from following through on the really bad ideas is the salient part here :) This holds, AFAICT, for translation too. Yes, you can get the gist via machine translation, for some language pairs. You cannot preserve imagery, intention, subtlety without a human int the loop. AI is good at looking things up and beats us there. It is entirely insufficient at context & wisdom (and I don't see that changing anytime soon). Sure, the industry might choose to collectively ignore that and jump off a cliff. It's not going to end well, so I hope saner heads prevail.

foobiekr 7y ago

A somewhat related story. 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.

groby_b 7y ago
UI design will live for a long, loooong time, because every single customer is basically like "I'm somewhat of an expert myself". Preventing them from following through on the really bad ideas is the salient part here :) This holds, AFAICT, for translation too. Yes, you can get the gist via machine translation, for some language pairs. You cannot preserve imagery, intention, subtlety without a human int the loop. AI is good at looking things up and beats us there. It is entirely insufficient at context & wisdom (and I don't see that changing anytime soon). Sure, the industry might choose to collectively ignore that and jump off a cliff. It's not going to end well, so I hope saner heads prevail.