3 ms·
"We tend to overestimate the effect of a technology in the short run and underestimate the effect in the long run." -- Roy Amara disclaimer: I have a Ph.D. in
by martincmartin 9y ago
"We tend to overestimate the effect of a technology in the short run and underestimate the effect in the long run." -- Roy Amara
disclaimer: I have a Ph.D. in Robotics from CMU.
AI is similar: in the 1960s, it seemed machine translation was "just around the corner," but it turned out to not be practical for a long time. Until it was, in the 2000s or so.
- Eridrus 9y agoI think it's worth noting that the approach to translation has changed significantly since the 1960s. AFAIK the real breakthrough wasn't that technology just got better, but that the whole paradigm shifted to statistical machine translation in the early 90s, and the whole paradigm is changing again from phrase-based statistical translation to neural network-based translations. The shift to neural nets can be somewhat attributed to increased compute capacity, but it's not an automatic result of it. I have no idea what the state of the art is in robotics, but I expect there will be some fundamental changes to how we do robotics before it becomes a break-out success.
- Retric 9y agoIt was really not reasonable to do this stuff using 60's erra computing technology. So, it's more new tech enabling new and less efficient approaches than generally talked about.
- Eridrus 9y agoThat's definitely true, but you need both, and having access to more compute doesn't automatically mean you will find new approaches, someone actually has to put in the hard work.
- Retric 9y agoTrue enough. But, to be clear many of these ideas are 20+ years old they simply did not work back then.