Blog Post

When our virtual assistant does not understand our intentions, but executes our instructions perfectly

To date, the effects that the large-scale implementation of artificial intelligence systems will have on our lives are unknown. We are getting used to having smart speakers in more and more homes that give us data and information on the weather, traffic and the results of our favorite team, and we are gradually adapting to them, just as a few years ago the OK Google or Apple’s Siri systems were introduced in our cell phones and computers and, with them, many people’s lives were made easier when it came to carrying out multiple tasks.

Opening the door to artificial intelligence

On the other hand, it has been clear for years that AI is here to stay at the consumer level. It is no longer only large companies that stand out for investing in R&D and for having access to cutting-edge technology that can boast of having computerized systems completely managed by one or more AIs, which make a large part of the operational decisions within the limits established for them by their programming and implementation, but, as more and more household appliances come standard with the minimum processing capabilities and connection to the internet cloud, they will be able to offer a series of functions that we now still see somewhat far away in time and somewhat related to science fiction.

However, the time is not so far away, for a part of society in the most industrialized and advanced countries, in which only when entering the house we say the name of our virtual assistant connected to the electrical system of our home, and we ask him to turn on the lights in the living room, to activate the heating and the temperature at which we want the room, and to put the oven on to start heating the food we bring for dinner. All this by voice, speaking “on the air” and the sensors installed in the central system where the home management AI is programmed, receiving the instructions and executing them immediately.

As this type of situation becomes more and more common, we can begin to give these central systems more responsibility and ability to control a greater number of household items. We may still dread the prospect of some household appliance making “decisions” on its own, even if they are decisions as simple as creating a list of something missing from the fridge or alerting us that the laundry has just finished and sending the signal for us to take it out and put it out to dry. This “central” AI, a sort of virtual butler, in the sense of integral management of most of the technological elements of the home, will be the one in charge of all those tasks of turning things off or on, sending instructions over the Internet to request something, or reminding us of an appointment if we have given it some access to the information in our agenda that, anyway, being in Google Calendar or any other similar system, is already under the same supervision of the Google AI that automatically adds us new tasks or reminders simply by monitoring our email inbox.

Giving instructions to “Vincent”

Few people, in general, realize that these scenarios we delineate are simply a product of human psychology for living more comfortably and with less hassle. Who doesn’t want to say “Vincent, turn on the bathroom light and turn on the shower so that the water is heating up, I’m coming in two minutes”? And the central computer at home, which we will have christened Vincent, will electronically give the relevant instructions so that our requests are fulfilled. It is simply a human desire for everything to be done as comfortably as possible and for there to be a minimum of effort, in some respects, to perform routine tasks. If something can be done without moving a muscle, we generally choose to create and implement the technology that can enable it.

Psychology aside, what is certain is that the introduction and development of artificial intelligence with its variants of Machine Learning, Deep Learning and artificial neural networks, together with the new computers and quantum processors that will multiply the power of our computer systems millions of times when they become available to the general public, will create a whole new model of life for a part of society. It is obvious that all this will not be implemented in the entire urban ecosystem or in all homes simultaneously, but their expansion will grow over the years, just as any new product does when it is accepted by consumers, and is allowed to expand to all corners and economies of the globe that have a certain capacity to be able to introduce them into their local social and economic ecosystems.

Your own personal “Jarvis”

New homes, for example, may begin to be built within a decade or less with intelligent home automation systems controlled by AI, with which we can communicate via cell phone if we are away from home, or, as we said, simply by speaking and voice commands once at home. It is an initial step for each person at home to have their own “Jarvis”, the name of the AI that Tony Stark, from the Avengers, had to manage his company and laboratory where Iron Man was “born”. Movies aside, this analogy and comparison is a bit what the technology industry has in mind for the next decades, an intelligent virtual assistant in every home, in every office and in every building, facilitating and managing everything that can be managed in them and freeing, somehow, people from small routines and tasks that are not particularly tedious, but also do not bring anyone anything special to do them, beyond the responsibility and control that comes from deciding for myself if I want something to be done this way and I do it myself, or simply telling Vincent to do it for me.

For the development of this AI, we are seeing how complex telematic assistance protocols are being developed initially through tests with chatbots, i.e. programs that can interact with the customers of a service or a company through its web pages and in which the chatbot itself solves, or tries to solve, most of the doubts, complaints and claims that the customer may have. The AI that resides behind the chatbot is limited to a certain set of parameters, solutions and answers, but that is only because it has not been programmed to keep learning and working on its own to incorporate more data and analyze more answers, with which to then build its own lexicon and its own ways of assisting in the claims or doubts that are raised.

Difficulties in some areas for AIs

Some systems, however, are trying to go even further. In some hospitals in California, a medical team is designing and training an AI to learn to recognize patients’ emotional states, since it is not the same for someone to come to the doctor’s office with a problem and explain it in a certain way, with the doctor understanding the hidden or transversal implications that the person may be suffering, as it is for a bot to simply analyze a few sentences and generate a diagnosis that may be more or less correct. The most illustrative case of this type of system is that of a patient who called the attending physician and connected with the chatbot that initially picked up the call. The message the chatbot received was “thank you very much for everything, goodbye”, which can be interpreted as a simple thank you message, and so it was processed by the AI at the time. But the doctor who heard the message was able to understand the emotional language and tone and realized that it was a suicide message, so he knew how to react and arrive in time to assist his patient.

And this type of situation is still far from the capabilities of most automated systems that we are implementing everywhere in virtually all areas of life and in all professional and service sectors, as it involves a knowledge of human psychology that we have not yet been able to program in our artificial intelligence systems, which, for the moment, are certainly limited to very specific areas, are certainly limited to very specific areas of knowledge or to question-answer functions within a data bank that, however large it may be, for the moment cannot encompass all the human instinctive knowledge about what is really being communicated or meant to us, when it is not being said to us in the direct words with which a robot or an intelligent machine would understand the message objectively, clearly and directly.

The new technologies that bring ever faster advances

But, in any case, it is really fast the work that is being done so that we have at our disposal these increasingly effective help systems and increasingly implemented “as standard” in almost everything we buy and that can be connected to the internet, through the “internet of things”, based on 5G technology and fiber optics to lead to the “cloud”, information and computing services hosted on “virtual” servers, spread across multiple locations on the planet, a whole technology of home automation assistance to which we are sure we will quickly welcome many of us as soon as we can see its operation, its response and its versatility.

We just have to keep control over those areas in which we are not yet willing to cede to computer systems with AI the interaction and problem solving, not only at the medical level as we have mentioned, which is the most obvious, but wherever human interaction is an added value for the resolution of critical situations, such as helplines for those who need to hear a warm and empathetic voice, customer service points where human presence is really important to solve compromising situations, as well as any element and area of our society where spontaneous human behavior is present beyond the simple routine question-and-answer interactions that an artificial intelligence can cope with, but not really understand, empathize with or be aware of, of what is going on in the life of the person who comes to ask for help and does not just want a manual of answers and automatic menus of “yes ask this then ask that”, but a broad and conscious understanding that encompasses much more than what can be programmed, at the moment, in an intelligent system, however much Jarvis or Vincent really do their role, their function and fulfill perfectly all their responsibilities.

Related Posts