My friend the fridge

RECENTLY, I HAD a series of unsatisfactory encounters with a disembodied voice. I was living in the United States at the time and I used to phone Australia using a pre-paid phone card from the US Postal Service. It was good value. I could chat for quite a long time for my $10, even if I had to endure, every time I dialled, an irritating ad for US Postal.

When my card ran out I found that the supply of phone cards in the shops was erratic. Sometimes you could buy global cards, sometimes you couldn't. But as it says on the back of the card, all you have to do to add value to the card is to ring the number supplied and follow instructions.

I tried to do this. Several times. I rang the number. A prerecorded voice gave me instructions. I followed orders. I keyed in my details as instructed. I wanted to add an extra $20. I keyed in my credit card number. I keyed in other numbers on the credit card. I keyed in just about every number in my life, until I was asked to key in the five-digit ZIP code of the address to which my credit card was billed. This was what stumped me. I am Australian. My credit card is billed to an Australian address with a four-digit postcode. Heart sinking, I keyed in four digits. The Voice told me it was not a five-digit number. I got smart. I added a zero to the beginning of my four-digit postcode and waited. Not for long. The Voice was endlessly patient with me. It had to be a five-digit number. I tried everything I could think of. I keyed in the five digits of my US address. I keyed in five numbers at random. The Voice knew better. After going round in a few closed loops, I was shot off at a tangent and told to ring back between the hours of nine to five, Monday to Friday, for a customer service operator.

Aha, I thought, I'll get the help I need. But that was not to be. I never reached a human. The Voice was never intended to divert a small spender like me to human help. I also tried asking the clerk at US Postal if she could add value with her activation machine but she couldn't do that for me. Whoever set up the system had judged human help to be unnecessary. What I wanted didn't come into the category of first-class service for first-class cards.

My time-wasting encounters with the Voice, in a call that "may be recorded or monitored as part of US Postal's ongoing commitment to customer service", is the way of the new world of machine-assisted living. I found no solace in the news that, in this series of encounters, there were other machines recording machines interacting with humans, so that one day, things might improve. There was a veritable community of machines out there, but not at my service. I'd been allowed a range of choices according to a predetermined pattern of question and response, but I was not allowed free will. No small spender of $20 should really expect more.

Washing machines save domestic labour in the family and, so far, have chosen not to talk to us. New generations of smart technologies promise (or threaten) more, as the family is extended to include a sub-family of networked machines. Synthetic voices, together with voice-recognition software, offer a means of communication without keyboards, and this may have some advantages. I want more. As machines start to talk to us, I want them to learn how to listen.

As devices designed separately and for different purposes become networked with each other, something new emerges from the system. A software agent exists to do the bidding of its human users, we are told by software engineers, though they are unable to predict the best interests of all future users. One thing is sure, I shall soon be interacting, on a daily basis, with all kinds of machines that will speak to me and interpret my responses. I shall be living in a smart environment complete with universal computing and ambient telepresence. It is in my interests to ask how the smart house will make for smarter family life. I want my efficiency to be increased, not just the efficiency of the corporations that provide the services.

I want to stay with the idea of the intersecting worlds of human and machine communities. I don't want to go down the cyberpunk path of considering the nature of the cyborg: the person into whom various electronic devices have been implanted, ranging from heart pacemakers to the trans-humanist fantasy of electronic brain implants. The trans-humanist vision strikes me as excessively individualistic. The trans-humanist seeks a personal state of total information awareness, together with virtual immortality if it ever becomes possible to upload the information in a brain to a computer. The cyborg manifesto is a vision of limitless personal power, with the individual at the centre of an austere cyberspace universe.

I am more interested in the notion of the individual as a member of both human and machine families. In particular, I am intrigued by one particular aspect of current promotion strategies. It's not the young, hip, cyberpunk enthusiast who is targeted in this new world order but the increasing numbers of the frail elderly. Imagine a networked world with the elderly person living alone at the centre of this network. This is a far cry from the cyberpunk world of cool young things who not only walk freely, with no evidence of arthritis, but who may also move at lightning speed in gun fights, those crafty implants in their heads giving them magical speed and dexterity in addition to total information awareness.

Back to the real world.

Why choose the frail older person living alone as the pioneer of the new smart living? The argument goes something like this: people now approaching late middle age are familiar with computers and mobile phones, and will pay for devices that promise to keep them independent in their own homes as long as possible. With the privilege of living longer comes the prospect of increasing frailty both in body and mind. By 2040 in the US, the number of people of age 65 or older per 1000 working-age people will increase from nineteen to 39. In Australia, the figures show a similar trend: by 2050 the increase is predicted to be from thirteen to 30. Marriage rates are declining, as is the birth rate, and a large proportion of older people will be living alone. This is happening, we are warned, at the same time that countries like the US and Australia face a severe shortage of nursing staff, especially in aged care. People are living longer. The birth rate is declining. Families often live apart, interstate or overseas. Who, or what, will do the caring in this future Australia?

My exchange with the US postal system was primitive. I had to key numbers into the phone pad to communicate. As speech-recognition and computer-vision systems improve, the keyboard becomes redundant and the environment becomes a "smart environment" packed with sensors that can monitor every move. Instead of interacting with a terminal, people will be able to interact with the space in which they live. The house will have processing power embedded in walls, refrigerators and even in clothes, and you'll be able to wave your arm at the wall to have a chat with an "embodied conversational agent". Increasingly, we will be surrounded by voices, both prerecorded and synthetic. I am clearly going to spend my old age, if I'm so blessed, interacting with devices that speak to me. I may find myself chatting to a sensitive wall. It's time to get seriously worried. New technologies promise practical aid, but will they deliver? In whose interests is this brave new world?

Children are used to toys that talk to them. Dolls say when they want their nappies changed or, if made for the US market, their diapers. I once saw a toy police car pulled into service by a two-year-old as a step to reach a high object. The car let loose with all its commands, bells and whistles on the go at once. I fear I may be about to be sold the oldie equivalent.

The technologies of second-generation telecare promise peace of mind, in part for the older person herself, but also for members of her family who may be unable to drop in every day. If I fall, who will know? Telepresence seeks to imitate the physical proximity of the distributed family.


IN OUR FAMILY, our 90-year-old great-aunt Nanace lived independently in her own home until the last two years of her life. She wore an alarm pendant round her neck in case of a fall and was supposed to press a red buzzer on a phone device each morning to let the monitoring service know she was up and about. Of course, there were times when she didn't wear the pendant alarm, or she pressed it to see if her support service would really spring into action (it did), and there were days when she forgot to press the red button, as indeed, I was told one day, do half the clientele of the monitoring service. Nance also had a small talking clock that she loved. As her sight declined, Yumi the talking clock became a reassuring companion. On the hour, every hour, it told her the time. It was a woman's voice, with a Japanese accent to its English. Great-Aunt Nance would often reply: "Thank you, Yumi." She loved hearing Yumi as she lay awake at night with the pain of her arthritis. After Nance died, Yumi continued speaking to me for well over a year, though banished to the laundry. I couldn't bring myself to put Yumi in the rubbish until her chip wore out.

Nance didn't live in a smart environment. She had stairs to cope with each day and she coped admirably with them, though it was alarming to watch. She had first-generation home telecare, the kind where the user gets blamed if the technology isn't used as the providers intended. A second-generation telecare is on its way, telecare with telepresence, designed to predict the user's needs.

The more I learn about this coming age of marvels, the more I relate it to Nance, to how it might have helped her or got in her way as she tried to live the difficult last two years of her life. Nance was of the generation where the technology of choice was the telephone, which she wielded with great enjoyment, even though, with her forgetfulness increasing, the younger generation resorted to answer-phones to avoid getting five messages about the same thing one after the next. The phone worked well for her, up to a point. Each morning her old circle of friends rang each other until, one by one, their voices stopped. Increasingly, we knew Nance sat in her home waiting for the phone to ring. We knew she was lonely, despite the best efforts of everyone. How might the coming second-generation of telepresence have given her more of what she wanted from life?

In an inquiry in October 2004 by the British House of Lords into the scientific aspects of ageing, the British computer scientist Peter Wright presented the notion of "lifestyle monitoring" in "second-generation telecare". The house system is "context aware", aware of who is in it and what that person is doing. As the resident walks around the space, she interacts with the system and communicates with it - for example, by waving at a wall. Once the daily routines are established, the system monitors any significant deviation from them. There may be computer vision sensors, pressure sensors in the floor, sound sensors on the water pipes and a fridge that is aware of what enters and leaves its door. The monitoring service records measurements of daily interactions by the user, so that any deviations from the everyday, for example, staying in bed for 22 hours or not using the toilet at all, can be picked up, and the house can ask you if there's anything wrong.

Take the fridge. Do I really want to have a conversation with my fridge? Well, I am beginning to suspect that one day it might come in handy. A talking fridge is not just there to remind me I am running out of food. It may become my virtual friend. In the House of Lords inquiry, Suresh Manandhar, an expert in artificial intelligence, spoke about the future of natural-language technology. The smart environment will have computational devices embedded in it, each small and specialised, but with power that comes with networking. So, an intelligent agent embedded in the refrigerator might be networked with computer vision sensors and might say: "It looks like you have fallen. If you do not move in the next 30 seconds I am going to ring the emergency services." The fridge cannot pick me up if I have fallen but it can call for help. And to have a fridge monitor its own contents is not an idle luxury. We know that at least once Great-Aunt Nance ate some ham that was decidedly dodgy and fell ill. After the event, we realised that ham had been in the fridge for about ten days. She couldn't see or smell that it was off. It would have been useful for the fridge to remind her that it had been there a while and perhaps she should eat it soon. Of course, this assumes everyone shops at a supermarket where all food comes prepacked and bar-coded, to be scanned as it enters and leaves the fridge. The fridge then knows what's coming and going. I think Nance would have been happy to have a virtual friend in her fridge, a second generation on from her talking clock, Yumi.

I think she'd also have liked a talking pillbox, especially one that could say, "Thank you for removing your pills for Wednesday morning. Have you noticed you've dropped the small red pill on the floor? How about leaving it there and taking another from the bottle on the shelf in the bathroom?" We were always finding tablets on the floor around her bed.


SOME EXAMPLES OF synthetic voices and personalities are found on the Cepstral website, Cepstral being a company that promised to bring "the power of speaking, reading and learning to life" with "natural, human-sounding voices". The voices are a lot of fun. Once voice, Millie, introduces herself: "Hello, my name is Millie. Some people say I look like Elizabeth Hurley, but I think it's just my voice." Yes, she does sound like Elizabeth Hurley. This could be the voice on my fridge. Another voice is Duchess, billed as "one of Cepstral's more sensitive voices". I was surprised to hear a man's voice. Was the Duchess in drag? The voices may be given additional effects, ranging from PVC pipe, dizzy droid, liquid love, to space-time echo. I suspect a humorist programmer enjoyed his work. These synthetic voices speak with smooth, natural-sounding phrases, with gaps between the phrases that make up the sentence. The listener knows it is a synthetic voice.

It could be that talking to my fridge may be a lot of fun. If the technology works as promised, that is.

The smart environment depends on programs monitoring other programs, programs learning as they go, programs that communicate with each other: all things that can go wrong. With first-generation telecare, most problems are attributed to the user, who is forgetful or who doesn't want authorities to know she has fallen, sometimes for good reason. Falls are one reason people have to move out of home. With second-generation telecare, the systems might be less easy to fool, and that's a worry. Sometimes systems are installed more for the peace of mind of members of a distributed family and I might prefer to keep my falls to myself, as long as I can manage to pick myself up again. What happens when things go wrong with the systems themselves – if, for example, the synthetic voices degrade and become unintelligible? Or worse. I definitely do not want a fridge with a demonic mind of its own, like in the movie Ghostbusters.

Nance spent much of her time remembering the dead. Her thoughts went way back to communities few of us knew. Her connections were not just with members of the distributed family as in geographically distributed, but those distributed in time, and many gone forever. As more film and audio records are taken by families, there will be more of this kind of evidence left of lives, and a picture-frame device that plays these sounds and images of the lives of past family and community would be something I'd like to have as an item in my future assistive technology.

How about system maintenance? Even if the systems will learn for themselves, getting the systems online and maintaining them will require skilled people. If the demographic trend is towards a decline in the number of working-age people, not only will there be fewer nursing assistants, there will be similar labour problems in system maintenance.

If there is to be a shortage of nursing help it will also be felt in retirement villages and nursing homes, and that's one rationale behind the development of Nursebot, a robot nursing assistant. Nursebot takes on the task of guiding residents round a retirement village. It can locate a specified person who has an appointment, remind him or her of it, ask if the person wants assistance, adjust its pace to less than five centimetres a minute if necessary. The researchers acknowledge that part of the function of a human nursing assistant is for conversation and Nursebot has some powers of communication, limited at present to commenting on the weather and what's on television. Perhaps some future Robonurse might be able to be programmed to listen empathically and to respond with a few noncommittal phrases, like, "Uh huh", or "Tell me more about it", or "I see", or "I'd like to hear more about that".


I STARTED WRITING about an unsatisfactory conversation with a prerecorded voice that got me nowhere in a commercial transaction. I've moved on to the problem that these voices, whether prerecorded human voices or synthetic voices, can only understand a few words because language, being social and cultural, is difficult to compute. I had dealings with a Voice that was, in my case I suspect, computed to be unhelpful, computed not to respond to my problem, interpreted as it was from non-US standard credit-card numbers, and asked a question to which I could not give a standard reply.

In the design of specific synthetic-voice devices intended as aids for the elderly, there will be instances in which they are going to be computed to be unhelpful. I can see an inhabitant of a retirement home telling Robonurse, "Take me home!" Everyone who still has some memory has a memory of "home", that wonderful place that was once theirs. "I want to go home" is a human statement. But Robonurse will be programmed not to take residents past the front door, that magical place where so many want to go. For Robonurse, home is somewhere outside its jurisdiction. For the person who wants to go home, it was once everything. Robonurse could be programmed to respond: "Home. Tell me more about it." And that, for that moment, may prove helpful.

Research for future products comes at a cost and it is assumed the user will pay. Some will be able to; many will not. Products envisaged for older people are also relevant for younger people with disabilities, but they are not a wealthy clientele. Researchers in robotics work at a far remove from the messiness of everyday life in a nursing home. Reporting from the International Symposium on Roboethics held in Italy in 2004, Bruce Sterling presented the worst-case scenario for robotic aids for the elderly: "The peripherals may be dizzingly clever gizmos from the likes of Sony and Honda, but the CPU is a human being: old, weak, vulnerable, pitifully limited, possibly senile." Sterling is being his usual flippant self. I wouldn't see us all as pitiable and senile, not just yet.

When I think more about the Voice, I start to wonder whether it really assists community or whether it may prove antagonistic. Assistive-living devices are designed to keep people "in the community" – meaning where they want to be, in their own homes in suburbs that they know. A community is something to which people feel, for whatever reason, a sense of belonging. Once it used to mean people who lived close to each other. Now it may mean people who may not live in geographical proximity but are defined by what they have in common – ethnicity, social class, gender, sexual orientation. Another thing people have in common, if they live long enough, is simply age. And that may be all they have in common. The nursing home is the object of horror, not only because ageing itself is an object of horror but because of the horror of being alone and helpless with strangers with whom all one may have in common is longevity.

In Justice and the Politics of Difference (Princeton, 1990), Iris Marion Young writes: "City life is composed of clusters of people with affinities - families, social-group networks, voluntary associations, neighbourhood networks, a vast array of small communities." What Young finds definitive in city living is also the notion of "being together with strangers". "City dwelling situates one's own identity and activity in relation to a horizon of a vast variety of other activity, and the awareness that this unknown, unfamiliar activity affects the conditions of one's own." When Nance could get up and about, she was an active part of her communities. Becoming more and more confined to home doesn't lessen the sense of belonging. It lessens the opportunity to contribute in previous ways and increases the sense of loneliness and isolation from community.

Ageing brings up the question of the loss of community, the loss of family, the loss of friends, the loss of the power to cope with everyday life. Assisted-communication devices address only some aspects of these losses, but may be enough to help for a while with the loss of mobility that affects social life. Enter technologies of the future. I want them to contribute to the many good things in life that endure, despite increasing frailty. Older people may interpret assistance differently from the young. In an article on justice and the ageing, Martha Holstein makes the point that elderly people can experience some forms of dependence and interdependence as "a positive source of renewed freedom and sustenance of the self". The Voice should be working to enhance the sustenance of the self.

We are at a point where worlds are colliding: the worlds of corporation and client; the worlds of human and artificial intelligence; the divide between the service the customer needs and the service the corporation intends to provide. It seems strange to talk about my future life as one that will be lived largely in the company of programmed devices with synthetic voices, but if these devices serve to reconnect me to family and community, then I'll be happy to embrace them.

What I don't want are irritating voice loops that take me where I don't want to go, or the infuriating loops within loops of mobile-phone voice-mail. I don't want to get stuck in auto-format mode, as I do from time to time, unintentionally, with Word. I don't want ads on my communication devices. I don't want mobile-phone ring tones on everything I own, reminding me to take my medicine or restock the fridge.

What I want may not be what I get.


WHAT I MAY get will be a community of programmed devices that will watch over me. They will talk to each other, and to me. That may be friendly. They may be programmed not to respond to certain of my wishes, if my wishes are deemed not to be in my best interests.

I want my smart environment to have colour, lights and action. I want it to amuse me. I want to be delighted with its antics. I want it to play music I can choose. I want lights that soothe or stimulate as required. I want to live with flocks of robo-birds, birds that move around me, never colliding. I want translucent robo-jellyfish that scuttle round on all eight tentacles, picking up the crumbs as I eat. I want bots to pick up my clothes for me when I throw them down, and dinner plates to scrape themselves and set off on tiny motorised wings into the dishwasher, through the cycle and out at the other end to the cupboard. I want shoes to shuffle themselves silently in pairs across the floor, self-polishing as they go. I want bots that have a solution for incontinence. For these I'll tolerate an intelligent toilet collecting my medical data and a fridge that gives me handy nutritional tips. No ads though. I want bots that skitter, bots that ripple, bots that love me, bots that respond to me.

I could get to like it, I reckon. But I want it all to work. 



Martha Holstein, "Opening New Spaces: Aging and the Millenium", Journal of Aging and Social Policy,
Vol 10 (1) 1998.

Websites, consulted January 31, 2005.

Cepstral: we build voices. Text to speech synthesis.

Compsim LLC: For devices that think.

Nursebot project: Robotic assistants for the elderly.

House of Lords. Minutes of Evidence taken before Science and Technology Committee. Scientific Aspects of Aging. November 24, 2004. Uncorrected transcript.

Bruce Sterling, "Robots and the rest of us", Wired,

Get the latest essay, memoir, reportage, fiction, poetry and more.

Subscribe to Griffith Review or purchase single editions here.

Griffith Review