Shelly, Shelley and Shelley Heights discussed on Chips with Everything
Turquoise like shape was the perfect choice to implement. This functions is friendly appearance as slow behavior are perfect for interaction targeting children, and also retracting motion can be implimented into restraining churches, abusive behaviors by even active. So what is it being used for now, you know, is it in schools? Is it just in your lab and children are visiting you? How is Shelly being used? Actually, as one of this lags Burkan after the final test. So it's all repair. So after repairing elegant displayed at the science museum in Korea. Oh, wow. Okay. How many children have met Shelley? The Shelly was made during the neighbor, less internships, and the Shelley met children in one session of monthly events held by neighbor less. So at this day, employees can bring their children to the headquarter. And their children commit Shelley play with jelly. So in every about one hundred to John came for the event every month. So what age group are you testing Shelly on or was Shelley tested on? Actually any children can play with Shelly. However, Ford, a research on harshly restrain, the children's visit behavior. Shelly was tested with children in six to thirteen. Okay. And what was the general reaction to Shelly? So most children are very curious about how Shirley react to their actions. So as Shelley highs, it's has unlimited inside some children like to hit Shelly. But at the same time, if Shelley heights is suspense, all is interactions. So until Phil unhappy about those abusing actions. So because they wanted change with Shelly. So any aunt Trajan generally tried not to Shelley. Even more they actively suspended or discouraged other children's abusing behaviors. So why do children need to learn NAS to abuse robots? Because there's no, that that many social robust real appear to our society soon. Right? So people need to know they need to treat well rowboat in your life. This is kind of too futuristic perspective, but we need to treat everything in a good manner. Not only for rowboats, buttocks. Affor- people are on us interacting with us. So Shelly was created to prepare kids for a future in which they will regularly interact with social robots. But it's not just humans who are doubting to this possible future. The is we interact with now all like Google assistant being designed to sound more and more like human beings because the tech giant's say, that's what uses prefer. After the break, we'll talk to a professor of spoken language processing who perceives this looming future of robots that sound just like us and doesn't like what he sees the shallowness to the knowledge unto their abilities. That create potentially creates a serious problem for users who might make totally the wrong assumptions about doing. We'll be right back. If it didn't kill people, it is a fantastic mineral. It is fantastic additive to any construction project that you're undertaking except that it does kill people. Us best us the wouldn't material that infiltrated buildings everywhere in Britain and beyond protecting infrastructure against fire, making stronger. It was the perfect solution for postwar contrary until it wasn't. Even health risks, special six Asia and now well, established countries continue to build with this material, but why the mantra from the specis corporations where it saves more people than it kills than so than it's worth the sacrifice worth the risk to find out more head over to the guardian dot com forward slash podcasts, research science, weekly, on your podcast up. Welcome back to chips with everything. I'm Jordan. Erika Weber. Before the break we spoke to Hinton goo about Shelley, the robotic tortoise that she team designed as an aide to teach children not to abuse their robot companions now or in the future. So we've heard about attempts to teach humans how to interact physically with robots that I wanted to know more about how he talked to them. So on professor Roger Moore, I'm a professor of spoken language processing here at the university of Sheffield, and I'm part of the speech and hearing research group that we have here, which is one of the large groups in the UK that works on because speech technology and these days, I'm very concerned with how we interact with machines using voice and how that is integrated with other ways that we might interact with machines. And particularly with robots is an aspect of what I'm looking just to the moment. So have you come across research about how humans speak to robots or artificial intelligence? Well, there's a huge amount of work and it's growing substantially because as you're probably were the underlying technology for speech basis, Dems. Just improved dramatically in the last few years. So it's not alone ago that you know if you spoke to a speech recognizer than it would probably get quite a large part of what you said wrong, but that is all changed. And so that's means that there's a whole new world opening up now where we can envisage speech based interaction with so-called intelligent artifacts. Hey, given that we can almost have conversations with robots. Now, what interests does that raise for academics to the dramatic improvements in the underlying technologies? So of the three, three main technologies involved, automatic speech recognition, then the the opposite to that is speech synthesis technology. So that's taking some representation that's inside the machine, turning it back into speech. So allowing the machine to talk back. And then the third sort of core technology is what we call on the lying dialogue manager. So all three of those technologies have improved dramatically, but there are still some serious questions about about each of them in the past. The way that machine spoke was brought the simplified or you know the the quality was not great, but you've probably aware that even, you know, in recent times, the voices of become very natural beginning to sign almost like. A natural human being. And so, but there are still a lot of research questions there about, well, yes, it can say if you things, but should it be more expressive? Should it have emotion? What about giving that put voice various accents, or styles or personalities, all of that is still being looked at, but there's a, there's a whole other set of questions about the total system you told better conversational interaction will do. We want to have conversation interactions with AUSSIE facts which are basically not another human being, but which purporting to be a will use his know what to do in that situation. We know how to talk to people, but do we know how to talk to a machine because these systems on not people, there's a, there's a shallowness to the knowledge and to their abilities, and that create potentially creates a serious problem for users who might make Tony Romo assumptions about what they're doing with. Is there any research that suggests that kids who grow up in writing with a is in this way, might then end up interacting with other humans in that way? Not little world. But I think it's all too early. You know, these things have only only only appeared in the last couple of years, and I'm gonna have to say the technologies running ahead of some rather important ethical questions here. One of my bug does the fact that's a the voices that are used in these agents. So all human like my view is that is totally inappropriate of the agent is not a human being, does not have the capacities, the beliefs, you know, the deep knowledge, all of that of a human being. And so by speaking with a human voice in, it's actually in some senses deception, this is a very unpopular view, but it would be easily possible to also the voice of one of these offficial agents so that it signs robotic. Why do you think it is so popular to have these robots sound like humans, talking to developers and the companies are putting these things that I get the message back. This is what. Uses want. I don't know if you saw the recent coverage of Google new duplex.