Recently a leading smartphone maker incorporated voice recognition and natural language processing in its latest product.
Somewhere, on the railways of new technologies, there is a train waiting for those who believe that voice based human machine interfaces are a natural unavoidable development. Apparently some smartphone makers believe that they must not miss this train or, perhaps, that they will be the first to really set this train in motion. In the meanwhile there is another train, which has already started moving around. Passengers in this second train have re-trained themselves and accepted the fact that machines seem to feel better with complex sequences of keystrokes rather than words. There is perhaps a strong belief among the so far few passengers in the first train that once they get it moving, they will soon overtake the second train.
Perhaps, and despite the fact that HAL ought to have appeared exactly ten years ago, I will probably live not long enough to see the final outcome of this race. What mainly concerns myself is the change in everyday life that will be brought by the introduction of new technologies in real products.
There is an advertisement showing the new fascinating capabilities of the aforementioned smartphone: A white-collar worker appears running in a park. He reschedules his appointments by using the new voice based human-machine interface. Clearly the advertisement is very carefully designed. He is alone; those who might overhear what he is saying cannot threaten his privacy. He is running; he cannot use a keyboard. He is not driving; he can afford to increase his degree of involvement in the “discussion” without endangering himself or others.
Now imagine yourself in a public space (bus, train, airplane, waiting room). You must whisper for the fear of being overheard. On the other hand, the more extrovert part of the population around you will probably not care, they will speak out their commands to their smartphones, and terrible noise will be created. Not only you will be annoyed by the addition of useless sound decibels, but also by the fact that you will not be able to whisper any more.
As a further example, imagine yourself on the road, driving among cars, whose drivers are prone to new kinds of misdemeanors. They can now violate the law not only by talking to their friends, but also by dictating letters and email messages to their smart-phone talk-bots. In addition, they will become more and more nervous as soon as they find out that they have to correct a talk-bot’s understanding of what they have just said or meant to have said.
Then there is the serious issue of context awareness technology maturity. Context awareness in particular and ambient intelligence in general, are well-researched areas. Ask your smartphone to remind you to collect your keys when you leave home. It will create a location dependent reminder to do that. However, the location is sensed with a certain (feasible or desirable) accuracy, and you will be reminded only once you are some hundreds of meters away from home, i.e. when you are already locked out of. The problem is not purely “technical”, it is also a problem of logic. How does you smart assistant understand that you will be leaving home while you are still there? A proper answer would probably rely on a combination of events, e.g. your calendar, your daily routine, and indications of movement from accelerometers hidden in your smart-briefcase or in your car keys. And still the real truth, i.e. the intention to leave home, lies deep inside your brain.
Perhaps the elderly will find in the currently naïve talk-bot of their smartphone a consoling voice, and kids will rediscover it once more as something between a toy and a friend. Perhaps using the more natural voice interface can eventually save some lives. Perhaps significant experience will be gained and the side effects will eventually be smoothed out. In the meanwhile we shall become part of yet another mass scale field test with or without our consent.
Somewhere, on the railways of new technologies, there is a train waiting for those who believe that voice based human machine interfaces are a natural unavoidable development. Apparently some smartphone makers believe that they must not miss this train or, perhaps, that they will be the first to really set this train in motion. In the meanwhile there is another train, which has already started moving around. Passengers in this second train have re-trained themselves and accepted the fact that machines seem to feel better with complex sequences of keystrokes rather than words. There is perhaps a strong belief among the so far few passengers in the first train that once they get it moving, they will soon overtake the second train.
Perhaps, and despite the fact that HAL ought to have appeared exactly ten years ago, I will probably live not long enough to see the final outcome of this race. What mainly concerns myself is the change in everyday life that will be brought by the introduction of new technologies in real products.
There is an advertisement showing the new fascinating capabilities of the aforementioned smartphone: A white-collar worker appears running in a park. He reschedules his appointments by using the new voice based human-machine interface. Clearly the advertisement is very carefully designed. He is alone; those who might overhear what he is saying cannot threaten his privacy. He is running; he cannot use a keyboard. He is not driving; he can afford to increase his degree of involvement in the “discussion” without endangering himself or others.
Now imagine yourself in a public space (bus, train, airplane, waiting room). You must whisper for the fear of being overheard. On the other hand, the more extrovert part of the population around you will probably not care, they will speak out their commands to their smartphones, and terrible noise will be created. Not only you will be annoyed by the addition of useless sound decibels, but also by the fact that you will not be able to whisper any more.
As a further example, imagine yourself on the road, driving among cars, whose drivers are prone to new kinds of misdemeanors. They can now violate the law not only by talking to their friends, but also by dictating letters and email messages to their smart-phone talk-bots. In addition, they will become more and more nervous as soon as they find out that they have to correct a talk-bot’s understanding of what they have just said or meant to have said.
Then there is the serious issue of context awareness technology maturity. Context awareness in particular and ambient intelligence in general, are well-researched areas. Ask your smartphone to remind you to collect your keys when you leave home. It will create a location dependent reminder to do that. However, the location is sensed with a certain (feasible or desirable) accuracy, and you will be reminded only once you are some hundreds of meters away from home, i.e. when you are already locked out of. The problem is not purely “technical”, it is also a problem of logic. How does you smart assistant understand that you will be leaving home while you are still there? A proper answer would probably rely on a combination of events, e.g. your calendar, your daily routine, and indications of movement from accelerometers hidden in your smart-briefcase or in your car keys. And still the real truth, i.e. the intention to leave home, lies deep inside your brain.
Perhaps the elderly will find in the currently naïve talk-bot of their smartphone a consoling voice, and kids will rediscover it once more as something between a toy and a friend. Perhaps using the more natural voice interface can eventually save some lives. Perhaps significant experience will be gained and the side effects will eventually be smoothed out. In the meanwhile we shall become part of yet another mass scale field test with or without our consent.
No comments:
Post a Comment