So why is AI back under the spotlight?
Well it helps that, tech giants Apple, Facebook, Microsoft and Google have decided to massively invest resources in AI as it is becoming the new buzzword in Silicon Valley. There is, however, some skepticism from the general public about the subject. Whether it is about the risk of jobs getting lost to the hands of robots and algorithms, this article won’t be about making the trial for or against the technology, but rather, try to shed some light on what we consider as an evolution of the world.
So here we are, at dawn of a worldwide use of AI. Although it has already been awhile since automated algorithms have been used by insurance and banking agents, the use of AI is going up a notch and is venturing in areas such as, the analysis of your personal data. This puts AI in the realm of understanding your habits and, above all, trying to determine what you need or want in real time.
So this is the positive side AI and, no need to worry, there are no glaring negative aspects to it. Advertisers will obviously profit from it by having even more targeted ad campaigns, but this is not a new phenomenon as most people are already knowingly sharing a lot of their information on the web.
What is interesting is the fact that tech giants are now investing a lot of resources in AI. Those companies will now directly participate in the evolution of AI, bringing some well needed capital to finance research projects which had lacked investment money for a long time. This is great news, especially in the light of the cloud and Big Data… AI really comes in handy for sorting, organizing and optimizing all that data… And of course, the hope for those companies and their shareholders is to squeeze a profit in the process!
AI is not new per se. Its constant evolution brought us the Deep Learning concept which has already been in use by your personal assistants Siri, Cortana and Google now for a few years. Those AI’s are based on what we call neural networks, which imitate how the human brain functions. But rest assured, the engineers didn’t dissect human brains in order to include neurons in computers! Neural networks are still based on semi-conductors and simply imitate how neurons connect to each other. As you can imagine, we are only at the beginning of the Deep Learning revolution. We can also expect a new slew of buzzwords along the next AI evolution wave… Bringing us ever closer to have a full simulation of the human brain… But as far as copying the human brain goes, AI is still ages away.
Do not fear!
We are already far more advanced as compared to the AI from the early 80’s. Those could be stored in only a few Kb but already were showing stunning results, mostly in single application domains at a time. One of the first AI application developed was for playing Chess. I still have a Kasparov Chess Computer, neatly stored in my mini IT closet museum. The AI at the heart of it only required 5 Kb of memory. To give you a better idea, that’s the space required for a color image of 40×40 pixels in size. As a comparison, 5 Kb is less than the memory required for today’s images used as icons on your desktop computers.
Today, algorithms are not bound to such memory restrictions. This order of magnitude is completely different when compared to internet and Big Data.. On top of it, everything is accessible from everywhere. It’s this general availability, anywhere at any time, which allows AI to be accessible on a large scale. Small businesses can now access those resources without having to invest large amount of money for their own IT infrastructure.
AI has made enormous progress. Watson, IBM’s flag bearer for its AI initiatives, is currently being used by thousands of businesses and research centers. Watson is flexible enough that its main algorithms can be applied to a variety of application domains. As an individual, you can also use it ! All you have to do is register to use their API’s, either on a test workbench or directly within your own applications. The service is also offered free of charge by IBM (given you remain within a limited number of monthly requests). You can register at the following url: https://www.ibm.com/account/us-en/signup/register.html
Other tech giants (mentioned above) also have their own AI offerings. For example, Facebook has ChatBots which allow its users to define a permanent dialogue “robot” which can be used for customer service requests 24 Hrs/7 days.
What about Robots ?
We have just used the term « Robot » in the text above. Here, at Friends of Buddy, we do not like using the term “Robots” when applied to algorithms. But what would a robot be without its AI ? Nothing more than a mixing robot.
The field of robotics has also made huge progress. We have now moved way ahead of the 80’s (that decade again !) which offered robots made of a few plastic parts (or wood, or metal for the talented do it yourselfers), a small motor, a couple of sensors and some Kb of Eprom memory used as storage for a rudimentary AI. Eproms were notoriously difficult to reprogram. Today, reprograming can be done wirelessly using Bluetooth.
The same concept now applies to robots which can nowadays access the internet, opening the door to even more possibilities. But let’s decipher what the true meaning of the word “robot” should be. Currently, the term “robot” is used to represent a hodgepodge of services. For example, those include Facebook’s ChatBots, autonomous vehicles, previous generation industrial robots, CoBots (collaborative robots), kitchen robots, vacuum cleaner and swipers robots, etc.. To that account, different types of robots get added to the mix. For example, should Echo, the “robot” from Amazon, also be added to the robot category? Echo is undoubtedly an AI within an object, but this object is barely just a cylinder equipped with a speaker, a microphone in addition to an internet connection. Echo can’t move and thus, cannot interact with its immediate physical environment. Same observation for the Jibo project, which by itself looks nice, as it can jiggle slightly, but has very limited impact on its immediate environment if just for the observer who can see it swerve on site. Similarly, a slew of IoT objects, with connected AI capabilities, could theoretically be added to the “robot” family. Should we then save the “robot” tag for humanoid robots? It’s not that simple to determine.. If we take Buddy, for example… It is not a humanoid robot, but certainly correspond to what we believe is a robot. Whether they are walking, ramping, rolling or even flying, new Form-Factors keep getting added to the mix. They all belong, potentially, to the extended family of “robots”. By trying to limit the term “robot” to humanoid robots, would be the equivalent of restricting the term « aliens » to human looking extra-terrestrial life forms.
Adjust the meaning of the term « Robot »?
Within the Friends of Buddy group, we often discuss the topic: whether there should be a new term created to designate the companion robots category. Buddy is a prime example (and, let’s be honest, if you are reading this article, you have a deep interest in Buddy !) as this applies to other devices as well. There will obviously be further evolutions of Buddy and even other robots created by Blue Frog in the future. In addition, other robots from different manufacturers and regions (Americas, Asia, Europe…) will inevitably get added to the mix. Also, more robots dedicated to specific tasks will be offered, along with more generalist robots in the near future. We could always, as we do today, just add a qualifier to better identify the robot. For example: Companion Robot, Assistant Robot and even a Comedian Robot. We, however, believe there must be something else, another term, especially knowing that “Robot” takes its origin from the Czech term “Robota” which means “Work”. Based on that origin, a companion robot would then be characterized more as a “slave”!
So what’s the fuss about ?
Well, to us, it’s pretty important. If we compare to more common objects around us, would we call a truck a “merchandise car”, a ship a “floating car” or even an aircraft a “flying car” ? This is the conundrum we would like to clarify.
Now, pretending we can find a new word that appliers to existing and FUTURE categories of robots would largely be pretentious, but we nonetheless would like to open the debate to the idea.
Robots for everyone!
There is a slew of new robots coming available early 2017. This marks a new era; this is the first time where robots make their introduction in people’s houses. Previously reserved to their builders, researchers and scientists, robots will now roam our homes while being configured to our liking. This is a paradigm shift for the robots creators as their “babies” will now end up in the public eye while they are helping families and incapacitated people. Have creators considered this new factor while designing their products ? For those who have and will in the future, the potential for profit is immense and we sure hope their guiding principles for the development of their AI’s will be done for the best of us all humans.
From brain to AI.
So what about AI. AI has also gone out of labs to enter our everyday life. It’s already in your house and even on yourself by the way of smart phones or sensors. No need to worry though as companies and researchers are focusing on user configurable AI’s. In best cases, there is even a desire to include a “humanity” function as part of them. This is crucial as, without it, we could imagine the possibility for the Sci-Fi doomsday scenarios to become a possibility. Many thinkers and philosophers expressed their concerns over that possibility and, this, provides our own guardrail from falling onto a cliff. We, humans, after all, seem to have the capacity to self-regulate as a way to protect ourselves! It’s in our genes!
In any case, AI is far from being able to replicate the magic that operates in a human brain, as we are just beginning to understand how the brain works.
In theory, taken gram for gram, computing power has now moved ahead of brain capacity from a raw computational capacity standpoint. But the real brain power relies on more than that, as AI researchers are also aware. Even with Big Data and neuronal networks, we are far from being able to mimic the creation power the brain has. If we just add the metaphysic or divine aspects of a human “soul”, the complexity of those processes makes us move in a circle, making it even difficult to discern the “beginning” from the “end” when trying to understand the process.
Like a Human?
Since the beginning of times, trying to duplicate what some believe to be a divine intervention, Human beings have tried to replicate themselves. This might have started with the myth of Pygmalion and Galatea, although in that case, it was a statue which came to life, thanks to the divine powers of Aphrodite. Closer to our current times, it is Leonardo Da Vinci who may have been the first to draw, and possibly build a humanoid robot which was an armored knight. The knight himself was not autonomous, but it was nonetheless functioning. Same concept in the pursuit of the ultimate AI: create a copy of the human brain which is as intelligent as a human being. But is this the right way to proceed? The idea behind trying to copy our brain is based on the fact that we don’t know any better. It is also possible that, although something better already exists, we may not have the capacity to detect it.
Conscience, thoughts, and sentiments…
Another important question is a source of debates : should AI’s and robots have their own conscience and sentiments (assuming AI’s can mimic our brains perfectly) ? This is where science and philosophy join together. A lot of contradicting theories have been elaborated, but one thing is for sure: in the quest of copying a human brain, we implicitly include the possibility for AI’s to have their own feelings, at a minimum, in the eye of their users.
It may be surprising to you, but we have unconsciously been preparing for this already. For example, as kids, we already were giving life to unanimated objects. We, then, were all talking to our favorite figurines, teddy bears and dolls. The evolutionary role could have been to talk to ourselves as a process of self-evolution. That process was of course in the total control of the user, which, in a way, may be similar to Google’s proposition to include a big on/off red button on each robot.
In a few decades, when robots will be as ubiquitous as cars and each family will own one or more robots, that question of an emergency shutoff will obviously be a very pertinent one. Especially if humanity elaborates a code of law, applying to robots!
Yes, this is the rooster chant, the symbol of France which is at the forefront of technology in AI and robotics worldwide. For example, Facebook hired Yann LeCun to head its research effort for their lab research center in AI in New-York.
Same thing for IBM Watson who promoted a French scientist as VP for Watson Core Technologies in 2012 who was previously at the head of Vivismo, a company IBM bought in 2000 which is leveraging Big Data to feed AI engines.
A few years ago, Sony dealt with Frédéric Kaplan who worked for a decade on AIBO, the first robot to make its way in our houses.
It is now your turn to bring your own ideas and ingenuity into tomorrow’s robots !
English translation by Jean-Pierre Lavoie