Sulphur on 15/10/2015 at 08:31
Sure. I mean audio recognition, not semantic/logic parsing. This is assuming that there hasn't been an initial voice familiarisation run for both programs though.
The 'bot is also apparently doing sight recognition, which is interesting, if not exactly discernible in the video.
Yakoob on 16/10/2015 at 05:54
Ah ok but we had voice recognition for a while now, like Siri as you mentioned. Even my phone can do a half decent speech to text job and that's with my accent.
I mean, yea when you think about it is kind of amazing we got technology to this level. But it's also no longer surpassing. Or maybe we're losing our sense of wonder and amazement ;p
Muzman on 21/10/2015 at 18:00
Jon Ronson met Bina once, with mixed results. Although they've probably improved her a bit since. Still, those videos don't impress that much.
(
http://www.radiolab.org/story/137470-everyone-has-solar/)
It sounds like 'she' does a lot of
stuff. What it amounts to seems fairly debatable. But someone's got to just dive in to this stuff really. So many things supposedly about strong AI or human-like consciousness is just philosophers noodling.
demagogue on 29/10/2015 at 05:02
I have the opposite reaction though. When I read about how much work goes into NLG natural language generation (NLG) and how intractable some problems are, then I'm pretty amazed we've come as far as we have with chatbots.
But then I'm even more awe struck at how causally humans do it, and find even the smallest parts of natural language amazing that other people hardly notice and think is nothing if they did.
It's like when you read a book on VR tech and look at an in-game tree for a while, then you go outside and look at a real tree. If anything, the sense of wonder is amplified IMO.
Edit. I guess I'm making an argument why people should keep their sense of wonder, but I'll admit many people are skeptical and not impressed even with genuinely amazing feats of technology.
Muzman on 29/10/2015 at 08:23
I don't know if that's the opposite reaction, even though I am fairly cynical about the likelihood of "strong AI" and human-like artificial intelligence.
It's more that lately every article on the subject is going to tell me about how it's scary because it'll turn us all into paperclips, or the ol' Chinese room makes it hard to tell what underlying processes are really at work yadda yadda.
Much prefer to see people just diving in and trying it out than theoreticians going back and forth (now I think about it, it's one of the few fields left, if you can call it a field- which is really the point I suppose- where theorists and philosophers do all of the talking. Once good solid technique really breaks out we won't feel like to talking to them so much. Which doesn't mean we should forget what they were saying, but, y'know. Attention spans and all that).
Language and mind seem almost intractable problems. So much ink has been spilled there though. But I find spatial recognition nearly as head spinning. There's been an interesting parallel development in robots doing autonomous navigation and the question of just how the heck we do it being hammered out in neuroscience.
I suspect we'll have wandering and fairly intelligent machines long before we can talk to them.
Ryan Smith on 29/10/2015 at 09:57
AI intelligence will never exceed human intelligence because we invented the former. We have brains we've yet to discover, which is why we seem stupid in comparison to an AI. There's an episode on MacGyver where MacGyver is faced with a battle of wits between himself and a super-smart computer which has control over an entire facility. The thing he beat the computer with was a gut instinct.
Pyrian on 29/10/2015 at 16:31
Quote Posted by Muzman
I suspect we'll have wandering and fairly intelligent machines long before we can talk to them.
I should hope so. At the end of the day, speech is an interface. Generally speaking chatbots simulate the interface while hardly even pretending to have any sort of intelligence
behind that interface.
Someone should make a natural language interface to X-com. No visuals. You have to talk them through encounters; describing what they should do based on their descriptions of what they see.
Quote Posted by Ryan Smith
AI intelligence will never exceed human intelligence because we invented the former.
So? What's so special about intelligence that we fundamentally cannot make robots do it better than we do? We have no problem making them stronger than we are.
Yakoob on 30/10/2015 at 02:16
Quote Posted by Ryan Smith
AI intelligence will never exceed human intelligence because we invented the former.
cars will never run faster than us because we invented them.
Oh wait...
henke on 26/3/2016 at 09:22
Hah, yeah. In other scary robot news:
[video=youtube;W0_DPi0PmF0]https://www.youtube.com/watch?v=W0_DPi0PmF0[/video]