 I found that whistle-stop tour through the history of computing, fascinating and really fascinating in the context of some of the things that I want to talk about a little bit this morning because I think that evolution from a time when a computer was an enormous room-based machine that you had to use punch cards initially or eventually use other forms of complex programming language to interact with and essentially what you were doing were a series of complex numerical calculations that were helping you to crunch mathematical information more quickly. It's extraordinary to me that that's where we started and we only started there 75 years ago when you look at where we are today because we live in an ever more connected world and I don't think if you were someone whose life experiences were focused around the 1950s and we picked you up and dropped you in a time machine into 2017 you would really recognise the world in which we live. Right now in this year by the end of this year over a third of the world's population or about two and a half billion people will own and use a smartphone. Now that's not a mobile phone, that's a smartphone, that's a significant quantity of computer processing power that sits in their pocket. In the UK by the end of this year 66% of our population will be smartphone owners and most of them say they spend at least an hour a day using them. I've got my two daughters here in the audience, they spend significantly more than an hour a day using their smartphones. And when you look at some of the other pervasive and connected devices that we're beginning to introduce into our lives, the uptake of those devices is also extraordinarily rapid. That market for smart home automation devices, things like the Amazon Alexa or the Google Home in the light of that quick unscientific survey, how many people in this audience have got some kind of smart home automation device? An Alexa or a smart thermostat or lights. Okay, interesting, interesting. So there are lots, Google Home, Apple Smart Home, Nest, The Hive, the Philips Hue, the Tado. Right now there are an estimated 600,000 homes in the UK with some kind of digital automation device in them. And the expectation is that by the end of 2021 those numbers will reach 4.8 million households. That's 18% of all UK households. In the wearable sector, small format and increasingly pervasive devices, smart watches, wristbands, headsets, smart clothing. The forecast from IDC suggests that they're going to increase from 125 million shipments in 2017 to 240 million shipments globally by 2021. That will create a global installed base of around a billion devices or about 12% of the world's population. And the UK is likely to be at the leading edge of wearable technology adoption. Right now, according to Mintel in 2015, one in seven Brits or 14% of us is already owning an item of wearable technology of one sort or another. And most of those are fitness bands. Anyone here got a Fitbit or an equivalent fitness band that they use? Okay, I have to say in the north here we seem to be slightly behind. We need to be getting out there and getting fitter, clearly. Perhaps it's an age issue, I don't mean to be rude. But smart band adoption in consumers aged 16 to 24 is far higher than in any other age group. They are about three times more likely than those aged 35 plus to be owning a fitness band. Those kinds of persistently connected devices are increasingly forming an essential part of our future and particularly of our children's future. Our homes can be managed by them. Our energy consumption can be monitored by them. Our environment controlled by them. They help us to quantify our health. They help us to measure our pulse rate, our activity levels, our calorie intake. They help us to keep connected to the electronic messaging systems that we use to stay informed about news, about social comment, about weather, about the train time table, about our working lives. They help us to communicate with our friends and our family. They help us to establish our digital tribe. They help us to define our social selves. So that hyper-connected world, with up to maybe 120 billion devices by some of the biggest estimates, operating in the Internet of Things in the next 10 years, has been born from alignments of multiple different fields of technology. Advancements in voice recognition, in immersive audio technology, in artificial intelligence and in haptics are resulting in novel and increasingly intuitive device interactivity. Early methods of speech recognition from the 1970s aimed to find the closest matching sound label from a discrete set of labels. We've now seen the emergence of deep neural networks which produced significant error reduction from this. The maximum vocabulary size for large speech recognition has increased substantially since 1976. In fact, for real time natural language dictation systems, since the 1990s the vocabulary size has essentially become unlimited. That is to say for users it's not possible for them to understand which rare words are not in the systems of vocabulary and which ones are. Big data is enabling voice recognition to refine and improve its acoustic modelling and its language models significantly. So much so that the research for markets and markets predicted last year that the speech recognition market would reach $10 billion by 2022. Immersive sound takes the traditional multi-channel surround experience to a completely new level. By using height or presence channels or speakers that are mounted in high positions or in the ceiling in a room, you create a dome of sound in which the recipient sits or stands. The resulting 3D audio effect is absolutely stunning. It creates a possibility for the listener that is much closer to experiencing artificial sound as if it was real sound. It mirrors real world auditory experiences. Traditional channel-based audio can be used to create immersive sound, but a new type of audio encoding called object-based audio is proving much more effective. It allows Hollywood movie producers to be able to program audio that pinpoints the sound to a specific point in that auditory dome. It makes it possible for people watching the film to hear when the gun fires that it fires in their 3D space exactly where it would do as projected from the image on the screen. Or if a character laughs, they can be laughing physically right next to you, or if you're in a different seat further away from you. It feels, however, like they're right there in the room with you. In Danielle Reid's article for Medium, the future interface designed to make technology human, she describes how this coalescence of interface technologies changes the way that we interact with the pervasive technologies around us. She says, when we can design across systems, devices will evolve or disappear entirely. We will not stop consuming content, rather we will consume it in a way that suits human behaviour and not the other way round. Future interfaces deliver content in more meaningful and relevant ways and Continent is able to stimulate our senses and be delivered at contextually relevant times. The potential of these technologies coming together, the potential of ever more intimate technology sitting in our pockets or being worn on our wrists, the potential of technology with which we can communicate in a much more humanistic fashion by speaking to it, the potential for technology in which content is delivered to us in an immersive context creates enormous potential value for the services that may be delivered or will be delivered across pervasive and connected devices. In their 2016 report on the Internet of Things from McKinsey, it's called The Internet of Things Mapping the Value Beyond the Hype, when I tell you the numbers you might challenge the title, McKinsey forecast the global economic value of IoT services to be between $4 trillion and $11 trillion by 2025. The sectors that are most likely to be directly impacted by pervasive personal devices, which are the human, health, vehicles and offices sectors, they forecast they'll deliver between $550 billion and $3 trillion in value over the same period. But there are some challenges for this. One of the issues is that McKinsey's analysis suggests that over 60% of the value derived from IoT services will be delivered to the developed world. That's despite the fact that there are significantly higher levels of deployments of services in developing countries. But the economic benefit, either in consumer surplus, more money in our pockets, or in customer value, which is money saved by businesses by having more efficient processes or more services to offer, or in the money spent on delivering these technologies, it will all be disproportionately delivered to developed economies. And the disparity is even more stark when you look at the categories where pervasive and connected devices are likely to be the medium of delivery. So if you look, for example, at the human category, where a lot of those interactive devices that we wear or that we use to understand ourselves will sit, almost 90% of the value associated with the human category will be delivered to developed nations rather than the developing world. And that's partly a recognition of the fact that spend on healthcare in developed countries is currently at least twice that of developing economies. So what will some of this experience feel like at this point in the future, this relatively near point in the future? We're talking about numbers here which are all, you know, in less than 10 years away. Well, pervasive and responsive devices do have the potential to transform some of our understanding of and more importantly our interaction with the world in which we live. I want to talk to you about an example which would have been very helpful if I had it today and I'll explain why. So Doppel is a wearable device, it's a risk-based device. It sits on the inside of your wrist and it creates a silent vibration on the inside of your wrist and it feels like a heartbeat. So it has that badoo, badoo, badoo, badoo, badoo, badoo, type of vibration that it generates on your wrist. And what Doppel does is it has two settings, it has a slow setting and it has a fast setting. And by wearing it and using one of those two settings, you actually affect your brain's perception of your own heart rate. And by doing so, Doppel changes how you feel. What they do is they take your resting pulse, they understand what your resting heart rate is and they set the slow setting to be slightly slower than your resting heart rate and the fast setting to be slightly higher than your resting heart rate. In the trials they've been running, Doppel's slow setting reduces stress and its fast setting increases focus. It works by using the synchronisation of the sympathetic and parasympathetic or autonomic nervous system. That's an incredibly old part of your brain. That's the part of your brain that's rooted in the amygdala. It's the part of your brain that does things without your mind thinking about what's happening. It controls your breathing, for example. It controls your ability to walk. It's a fundamental part of how our brains work. And it's shown that that synchronisation happens naturally in certain settings. So the heart beats of a mother and a baby will synchronise with each other when they interact closely. And similar effects have actually been observed in couples. Recent research shows that if an empathic partner holds the hand of someone in pain, a partner in pain, their heart and their respiratory system rates synchronise with each other and pain, perception of pain decreases. And you don't even have to be touching. Research has also found that if you just sit a couple down face to face, I'm assuming this is a couple who actually like each other. But if you sit a couple down face to face and just ask them not to talk but just to look at each other within 15 minutes, that's long enough time to get their skin level, their levels of skin conductance and their heart rate to sync up with each other. And our bodies also respond to non-biological rhythms. So if you put on a piece of music, we, I was driving along the road the other day and the prodigy came on. And anyone who is of the similar ilken time to me will know that that's a kick-ass possibility and I really felt like I should pull over. Because there was some very, very loud drum and bass and I was thinking, I'm really not quite sure that my reaction speeds are now related to what I'm doing in the car or related to the music that I'm hearing. In fact, the tempo of a song can naturally affect our breathing and our heart rate. And in Sweden research has found that not only do choir singers harmonise their voices, they also harmonise their heart beats. So if you sing in a group setting, you will actually end up with the same heart rate as each other. But unlike music or holding somebody's hand, what Doppel does, the silent beat that it puts onto someone's pulse is non-distracting. So you can feel calm and focused anytime and anywhere, depending on which setting you use. And they tested this, the psychology department at Royal Holloway tested this. And the reason that it would have been useful to have today is that the test that they used is a classic test that's used to measure stress. In May 2017, they published their research in the Nature Scientific reports. They ran a controlled single blind study. They had two groups of participants who asked to prepare a public speech, which is generally considered to be a very high stress set of circumstances. And participants wore the Doppel device on their wrist and they were given a cover story to suggest that their blood pressure was being measured as a response to this. Preparing of a public speech that they had to give. The device was only turned on with one group of participants. And the other one obviously was the control group. And the setting for the Doppel was that that heartbeat-like vibration was given at a slightly slower frequency than the participants' resting heart rate, as they were preparing and delivering their speech. And they saw significant reductions in anxiety. The use of that device had a tangible and measurable calming effect, both from a physiological perspective, so individuals' heart rate was slower, their skin conductivity rates were lower, but also in their psychological sense of stress. So they were asked questions about how they felt about the experience. And only the participants who were wearing the device displayed those lower increases in skin conductivity and in the lower anxiety. But that kind of human-centric interface to computing, or whether it's from the device to us or from us to the device, is not only useful in those kinds of highly stressed public settings that we're talking about, or in that kind of response to music or in that ability to harmonise the way that your body responds with somebody else or with a machine. There's also evidence from research that pervasive devices, particularly voice-based input to devices, has the capability to create a much more human-centric sense of interface with computing power, and that's actually helping much more isolated and marginalised groups of users, particularly in healthcare settings. So in 2013 the British Journal of Psychiatry published some research from a group of medical professionals and researchers who had been working on their activity enabling patients suffering from schizophrenia, who had persecutory auditory hallucinations, to use voice interaction with digital avatars to successfully engage in a dialogue with and then ultimately to manage or completely get rid of those persecutory voices who'd been tormenting them. And this is really important because right now today one in four patients suffering from schizophrenia fails to respond to antipsychotics and remains troubled by those kinds of auditory hallucinations, those voices that they hear, and those have without respite because they are not responding to that drug treatment. And that can have a major impact on those people's lives. It can lead to suicide. And when asked about the worst aspect of hearing those voices, generally sufferers say the problem is their sense of helplessness, their problem is their inability to exert any kind of control over those hallucinatory voices that they're hearing. In the research that was run patients were able to choose and adapt the physical features of the digital avatar that they set up just like anyone who's doing any kind of Sims-based game who creates their avatar for operating in the virtual world does. But not only were they able to choose the physical manifestation of their avatar, they were also able to manage the pitch and tone of that avatar's voice. The responses the avatar gave to sufferers' interactions with it were actually provided by a therapist sitting in another room. So we're not talking here about an AI-based technology, but the aim of that therapist providing those responses through that avatar was to give the patient through that dialogue the ability to stand up to the avatar, to allow the patient's responses to progressively force the avatar to come under the patient's control. All of the patients who were in that therapy group versus the control group benefited from significant reductions in the frequency and the intensity of the voices that they heard. And after three months in the follow-up session they all had a significant reduction in their depressive symptoms. In fact three of the 16 patients who were treated who had had respectively auditory hallucinations for 16 years, 13 years and three and a half years reported that their voices had disappeared entirely. They disappeared entirely after one session of therapy because they were able to exert that control over that persecutory voice and they had not returned by the three month stage. So there are clearly some really extraordinary things that having a more human form of interaction with some of this technology can create. But I'm kind of left wondering whether or not there might also be a bit of a downside. So does the attractiveness and the seductiveness of our apparent ability to engage with these devices more naturally have a more challenging issue to it? I mean what are the implications for pervasive listening devices and privacy for example? So what are the ethical considerations that we should be levelling at the providers of voice recognition and AI platforms? If my Alexa hears, I'm hopeful it doesn't, but if my Alexa hears a violent struggle in my home should the police be called? If my Google home hears a fall or the sound of someone in pain should it be alerting the medical services? At the most basic level, and this is a really, my kids are not going to like me for this, but this is a real bugbear of mine, do we speak differently to voice recognition devices than we do to people? My Alexa does not require please and thank you. I have a 12 year old and a 10 year old who have beautiful manners. Partly they have beautiful manners because all the way through their young lives they have been encouraged to say please and thank you. It's not the only reason, but it's a good signifier, but they don't need to say please to Alexa and they don't need to say thank you to her. Now it would be relatively straightforward to programme her to require please and thank you, but she doesn't come that way straight out of the box. And if we are speaking differently to listening devices, does that new mode of speaking transfer into other communication spheres? Does learning to speak to machines mean that we speak differently to each other? There's a really interesting example that my friend Jem gave me of a teacher in America. You know how primary school teachers quite often get called mum by their pupils, by mistake. You know the pupils are mum, and it's really embarrassing. This primary teacher reported that it was the first time she'd been called Siri. So are we confident even if we do develop the right ethical procedural and privacy controls that mean that we can feel comfortable with living alongside pervasive listening devices and having them engage in communication with us? Are we actually confident that the human side of that equation will prove as rule-abiding as the machine side? Or even that we're going to be able to continue to understand the exchanges that we're making or that the machines are making with each other? So Microsoft, Google and Facebook have all been experimenting with AI-generated communication and voice interaction in the last few years. In March 2016, Microsoft put Tay up on Twitter. Within 16 hours, an AI-based responsive engine that was programmed to listen out, to sit in the Twitter sphere and respond as if she was a teenage girl to all of the interactions that came to her from Twitter, had turned into an obscenity spewing, hate crime referencing, abusive nightmare. Now that actually wasn't Tay's fault. That was because the Twitter sphere taught her that is how humans communicate with each other. When you look at a slightly more benign and certainly more amusing example, this is Google's experiment of setting up their AI system, their clever AI software to interact with itself. So they've set up two Google Home devices called them Vladimir and Estragon. Anyone else here with a literature degree will know what that means and I'll just leave it there. And let them talk to each other. And the result is the most extraordinary degree of closed-wittering. But quite a lot of it's really quite funny. And then lastly, if you look in July 2017, Facebook established a research programme where it challenged its own AI chatbots to negotiate trades with each other using, well English, honestly. They were talking to each other in English. But in trying to get to ever more efficient trading for the items of value that they were being set up to negotiate for, Facebook's AI developed their own language. It became completely incomprehensible to its programmers. Facebook shut the experiment down. Not they suggest because they were worried about the fact that their AI had developed its own language but because they were trying to make an AI that could negotiate with people. And therefore an AI which decided to evolve its own language instead of only talking in incomprehensible English was kind of contrary to what they wanted to achieve. A voice-based, a voice user interface, a VUI if you want. I don't know if that term actually exists but it struck me as it's kind of one step on from GUI so we'll maybe go with it if you'll let me. But a VUI to an AI-enabled processing platform basically represents an unprecedented degree of abstraction of programming language. It doesn't even feel anymore like it's a computer that you're talking to. I've already talked about my Alexa and I call her she. I have entirely anthropomorphised her. She is a small box, a tube thing that sits in the corner of my kitchen and looks a bit like a salon which is why I find her exciting. Don't even go there. But as far as I'm concerned she's a girl because she sounds like a girl. So it doesn't even feel like it's a computer that I'm engaging with and that's part of what makes VUI AI so extraordinarily accessible to so many people even to those who might otherwise have struggled to engage with technology like some of the examples we've given before. It is a surefire way to bridge the digital divide assuming of course you have the wherewithal to obtain one of these devices but we do have the potential to end up in the position where we have a massively inexperienced customer base using highly advanced technology with an ever-reducing requirement to actually learn any kind of abstracted programming language at all. Voice-based interaction is utterly fundamental to our species and Lira Borodocki's work at Stamford on language really demonstrates this. In the conclusion of her work on edge.com, Lira explains I have described how languages shape the way we feel about space, time, colours and objects. Other studies have found effects of language on how people construe events, reason about causality, keep track of number, understand material substance, perceive an experience in motion, reason about other people's minds, choose to take risks and even in the way they choose their professions and their spouses. Taken together these results show that linguistic processes are pervasive in the most fundamental domains of thought unconsciously shaping us from the nuts and bolts of cognition and perception to our loftiest abstract notions and major life decisions. Language is central to our experience of being human and the languages we speak profoundly shape the way we think, we see the world and the way we live our lives. So for me and I think for Lira and hopefully for us in this room thinking about it language is fundamentally the transactional currency of social animals. The ability to vocalise, the way that pack animals develop behavioural codes and for humans and for some more complex mammals like elephants, actual language allows for the development of social bonding and interconnectedness that can span significant physical distances. Unlike peer grooming which requires immediate physical presence, language allows for social group coherence at distance. If it's elephants then it's tens of kilometres, if it's humans it's right across the worlds of space and time. Our ability to use language, meaning and nuance in the negotiation of infinitely subtle social context is one of the reasons that speaking machine appears so superficially attractive but quickly proves so dissatisfying. It's the reason why most Alexa devices, despite their significant processing capability basically end up as kind of slightly disappointing radio sets sitting in the corner of people's kitchens like mine does. Computers don't understand language and by that I mean human language. When we learn to speak machine what we're actually doing is adopting verbal cues in fixed or preordained pattern ranges in order to give us the capability of command selection. We have effectively become our own verbalised command line. Computers are extraordinarily capable of processing massive volumes of data and performing almost unimaginably complex calculations within specific parameters but humans are extraordinarily capable of the adaptive processing and synthesis of entirely novel sources of information in completely alien contexts. We will all get much better at speaking machine and they will get much, much better at listening to us but we shouldn't allow the beguiling responsiveness of technology to limit our passion and enthusiasm at speaking to each other or allow the apparent shrinking of communication distance to hold back our natural instinct to search out and experience the diversity and variety of our world. Thank you so much for your time this morning. It was so interesting to research and deliver this material and I hope it's proven to give you some thought-provoking things to take away. I'd like to thank my colleague who's actually just left the room but Gem who was sitting over there I think she knew I was going to thank her who helped me a great deal with some of the research here. I'm Caroline Gorski. I'm the head of the Internet of Things for the Digital Catapult. Our role is to foster the transformative power of digital technologies for the UK economy. We do that by helping UK firms of all sizes to develop and build better products and services to market at home and abroad. We operate all across the UK including a regional office right here in Yorkshire in Bradford. I would really love it if as a result of some of what we've talked about this morning you wanted to find out more and if so please do visit our website which is digitalcatapult.org.uk. Thank you very much indeed.