Video: Soul Machines: Humanizing the Interface Between Man and Machines
Andrew work speaks to Phil Chen, executive chairman of Soul Machines, at the Emtech Hong Kong Event hosted by MIT Technology Review and Koelnmesse. Soul Machines’ objective is to humanize the interface between man and machines, creating an emotional bond to our daily tools.
NEXCHANGE: Hi I’m Andrew Work for NexChange. I’m down at the EmTech event in Hong Kong and if you want to know the EmTech event is all about in one word, ‘geniuses’. We’re hearing about some amazing things today including a big session on artificial intelligence.
I am sitting here with Phil Chen. He is the executive chairman of Soul Machines. Horizon Ventures has invested into this New Zealand company and he gave a great talk today about what this company is doing and a totally different way about thinking about artificial intelligence and how it can be built. And Phil thank you very much for talking to us today. You kicked off today talking about looking at a very philosophical bend and talking about a philosophical divide that maybe people weren’t conscious of maybe until, for some of us, until you brought at the stage today. Could you tell us a little bit about that?
PHIL CHEN: Yeah well I was referring to a recent David Brooks talk. David Brooks is a New York Times Columnist and he was speaking of a formulation that were full of philosophical mistake that we’re making as a culture. And to me, it was the same I think philosophical mistake in A.I that we’ve been making and so what he said was as a culture, we’ve chosen Rene Descartes over Augustine.
NEXCHANGE: Okay and for people that don’t know Descartes and Augustine, what does that mean? So Rene Descarte, he’s usually thought of as sort of the father of the enlightenment.
PHIL CHEN: He is the one that says I think therefore I am and he sort of launched a whole trajectory of thought that people call Cartesian which is to say that you think of human beings as primarily cognitive creatures. And on the flip side you have somebody like Augustine. He was a theologian. He was a philosopher as well. But he was seen as a Christian philosopher and theologian and he would wax a lot about things like worship and things like emotional longing. And there’s a whole tradition of thought even that the whole that followed Cartesian thought in the Romantic period where that was revived. Where there need to be more emotion, the need to be more longing, it neede to be more human rather than seeing people as being reduced to just cognitive machines.
‘But the face has been the most natural interface for us. We’ve evolved into learning how to interpret and read faces.’
NEXCHANGE: Yeah and when it comes to A.I., how is that reflected in the split in thinking that you see an AI in the new direction that soul machines is taking?
PHIL CHEN: So what we see in A.I. in my talk, I say most all of the investments, most all of the media coverage has been in the cognitive side AI. Anywhere from reasoning to perception to natural language processing to planning, all these are very cognitive and I use the phrase ‘AI State of Mind’ purposefully. But what hasn’t been done in AI a lot is study in emotional feedback and emotional intelligence, face-to-face conversations. How do you relate to machines in emotional way? How does the machines read your intention just through your facial feedback, just through your tone of voice, just the way you breathe, the way you look at me, the way you nod? All of these tell you something about the person’s emotional state of mind that largely is ignored in a lot of AI design because in most of the devices that we interact with machines right now, it’s often disembodied. It’s not transparent. Even the voice controls that we are using right now, the cameras that we use. We don’t need that type or we have to train ourselves to learn how to use these machines. But the face has been the most natural interface for us. We’ve evolved into learning how to interpret and read faces.
NEXCHANGE: Soul Machines are developing these faces, kind of emotional avatars, that use the face as a model of communication. How is the computing different when you talk about this philosophical split? How is Soul Machines doing the computing differently to develop this new way of thinking about AI, this new way of taking action?
PHIL CHEN: Well I would say that it takes a more holistic approach. It’s not a reductionistic approach so it takes models of face-to-face interaction. It takes models of developmental psychology but Soul Machines has really built what we call the ‘nervous system’, so in our presentation you can see you can connect Siri which is right now the speech process. You can connect a different brain into this nervous system. You can connect different sensors. Of course you can connect cameras, biometric data that feeds into the system. But our system is focused on interpreting that data and we’ve built [all the philosophical virtues from Europe], from biological systems so I give the example of breathing. We simulate breathing because the way you breathe tells you about your emotional state. And every human being that speaks is dependent on the way they breathe. So to simulate a very realistic text-to-speech algorithm, for example, we need to simulate breathing first and that’s a biological model that we build first before we lay on top of the natural language text to speech algorithms.
NEXCHANGE: Yeah that brings us to another old divide that the miaphysites or the diaphysis on the mind and body separate. You’re trying to build an artificial body to go with your artificial mind recognizing they influence each other.
PHIL CHEN: Yes so in that sense, we’re very much again all very holistic. We do believe that you can think with your hands. You think on your feet differently from sitting down. I think we take that all into account when we’re building this autonomous character, this katana, this avatar.
NEXCHANGE: Okay, I’ve seen some of the work you’re doing and the avatar is on screen and it’s exciting, it’s pretty amazing, it’s a little eerie, it’s a little too real. If you like I’m going to change gears a little and ask, you came to Soul Machines as their executive chairman from Horizon Ventures. This is an investment, what was it about this that attracted Horizon Ventures as an investment and where do you think there’s going to be a payoff for this? I know that sometimes Horizon Ventures makes long-term bets. They’ll put money into something like when they got into Facebook in very early days when nobody knew what it was going to be or companies like Waze. Is this a long-term play where you don’t really know where it’s going to go or do you have a clear horizon for why you made this investment?
PHIL CHEN: Well I think at least my view in AI, if you look at all the big advancements from speech recognition and image recognition to planning navigation. A lot of it is not because we are smarter. It wasn’t better algorithms. Because if you look at all those breakthroughs, those algorithms with those breakthroughs on average has been invented 18 years ago. But the data that has been made available has only been three years ago. And that tells me that we haven’t gotten smarter. We’ve gotten bigger, cheaper and faster, but with Soul Machines, we’re trying to get smarter on the emotional computation side and it’s also gathering a completely different class of data that nobody’s gathering right now. There was Facebook where they get your social graph. That’s the type of data. Of course advertisers are interested in very different data, psycho-graphics data. But this emotional data has been avoided there and because of that we see an opportunity of this EQ data being a huge part of and potential for growth
NEXCHANGE: Okay so in the old days people were saying more of the old data. you’re looking at a whole new set of data and analyzing it in a different way. Alright, well Phil that’s great. this is really exciting stuff you’re doing. I know people going to keep a close eye on Soul Machines. I know as the executive chairman you’re going to keep a close eye on them and be interested to see where you go with that.