Beyond Artificial Intelligence: Hybrid Intelligence Systems.

(Spanish version soon).

1st contact of the 3rd kind.

AI, defined as something done by machines, is the actual hype (2019), the next hype is going to be the hybrid intelligence: A system that is compound by humans and machines which might have the next characteristics:

+ The Artificial part of it learns from the human.

+ The human part of it learns from the machine part.

+ The result of the system is unobtainable by the human nor the machine by itself. This means that knowledge is created by synergy and within the system and it is stored also within the system.

Observe that I referred to knowledge and not to information.

Clarification: “Stored” is normally understood in computation as “set in a physical substrate that is held in a geographical location” but in this context, we have to use a higher philosophical level: Any entity, in order to exist, must have an effect on other things. This perfects the “stored” term from merely “a location” to mean “in an actionable state”, for knowledge to be valuable it must be also applicable/actionable/practical. (In contrast with mere information which might not be used for any arbitrary lapse of time and just “sit stored” in a data center). And in this hybrid system, it is not actionable by any part of the system but by the system as a whole. Hence, in the hybrid systems, the knowledge is stored within the system in a way that only through the cooperation of its part this knowledge can actually be actioned (towards practical use).

I devised this technology as far as in 2012 when Louis Von-Ahn launched his public challenge after his success with CAPTCHA and other wide-use systems that learned from humans. Those technologies lead what AI does now today: Improved OCR capabilities and interpreting images tagging them with words. Readily, at that year, I designed the very first system of its kind that incorporated a specific type of dynamic: ‘Differential equations’, a type of which the most difficult ones are resolved by a method called ‘convolution’. Convolution is a method and a technique of mathematics that lets resolve differential equations, which in turn describes the most common way the living nature carries its processes and also how complex physics phenomena can be succinctly described.

It is not necessary to understand differential equations nor convolute methods to see the logic behind the previous statement (any phylosopher, or smart person for the matter, can): If a system is compound by machines and humans, and its complexity is designed to resolve very complex situations, then it should be constructed inherently at the height of the complexity of what it is aimed to solve. Thus, differential equations should be part of it.

The previous statement is my practical application of Einstein’s famous quote: “No problem can be solved from the same level of consciousness that created it.” Although “consciousness” has a deeper meaning, the logic inside the idea of that quote is absolutely (or relatively) the same I used in my “solution”! (joke! …but seriously!)

By nature, the logics are reversible and this lets one to find a practical solution (a “YES”) from a negative statement rule (a “NO”) simply by negating the premises. (Examples. “NO one is AWAKE = EVERYONE is SLEEPING”. “NO one is NOT awake = EVERYONE is AWAKE”. )

In an even higher philosophical approach: If a problem or phenomena has not been resolved or described by humans nor machines, then we can assume its complexity is beyond the common capabilities of humans and also of machines or it is beyond the methods the humans and machines are using.

Any equation is a short description of a system. And there are simple equations that are simple enough to be a workable tool for certain problems and phenomena. But for very complex phenomena, more sophisticated equations are needed. John Forbes Nash (1918–2015), laureated with the Nobel Prize in 1994 for his work in 1950(portraited in the movie “A Beautiful Mind” played by Rusell Crowe), and awarded the even more prestigious Abel Prize in 2015, had made it unquestionable. Apparently intuitive situations, seem complex, just because they are not modeled in the right way. Most of the phenomena in life must be modeled in differential equations. And solving them is not a trivial task but the opposite.

Complex phenomena that affects us all and which each and everyone of us should be concerned, like human politics, the functioning of the brain, intercultural communication, or even apparently simpler ones but highly “complicated” as the emotional mutual influence during communications, comprehending a spoken paragraph and its underlying intention, for all of them the equation might be very complex as it takes in account not just a large quantity of variables but because it is an ever-changing dynamic system.

We all ARE that type of system. When we hear some news, we might react emotionally, start to think in something related to that news, and then decide to keep that track of thought or to go back to what we were doing, perhaps taking into account the new information we had just received. (We decide if what we heard is information or we discard it as an interruption). So, yes, we are indeed ever-changing dynamic systems. Could a machine predict our behavior? Not really, it just can guess a good probability but it cannot find a deterministic future as either one of two: It does not exist, our minds are complex enough so that the system that is reading us cannot acquire the degree of complexity and inner knowledge required to meet the level of our behavior.

The inevitable conclusion is that in order to a system to be able to “absorb” knowledge and “hold” the amount of complexity of phenomena that humans involve, then at its structural level it needs to be at least of the same complexity and capabilities of humans. No computer does that (2019). And because of the architecture of actual computing technology, they cannot nor will. The best approximation so far is Big Data and it does not solve the equation of a single human, it just approximates to masses and averages or clusters of similar humans.

So if we want to understand a really complex phenomenon, and even try to solve really complex problems, then our approach, our tool, should have at least the degree of complexity and most importantly the type of behavior of that phenomena/problem. And in our case, <<human creatures in a physical world>>, that must include differential equations. Thus, two steps beyond that is convolution. Then, Artificial Intelligence is not enough, Hybrid Intelligence is needed.

Perhaps we have always had this problem, an erroneous approach to the things we most matter, and we had to create nowadays computer systems to try to reach our most desired goals. But I can see that we do not need extra computing power, nor memory, nor speed to solve them. The binary architecture in the overall chip circuitry (Intel) and our mainstream computational technology has gone beyond the point necessary for our needs. For our true needs. It has gone in an erroneous way in regards to “humanity’s goals”. It is useful for purposes we did not even imagine, like fake videos, virtual reality, and cinematographical visual effects. But those do not tackle our initial problems, our vision about things to resolve in life and society. I see clearly that with much less computing power, but a well crafted Hybrid Intelligence System, many phenomena can be understood and our most concerning problems can be resolved.

Actually, doing that, we will be opening the door to a radically new era. True cyborgs, I am talking about here. The singularity has its ways through the track I had just explained. Elon Musk’s project: Nueralink (to connect a brain to the internet) is just a tool to an almost undescribable situation where Hybrid Intelligence will harness power and ultimately take control of the new dimension. (Revision: Who said the singularity will bring one new dimension? I think it is going to be a Big Bang of Dimensions, The Next Big Bang).

We have already an Hybrid Intelligence system going on in SynerGears. We study how humans collaborate and work better through human consultants, which in turn pour their knowledge into machines, which help them to learn more about what they do, which is: To make people happier at work. Isn’t it a great job?? 😀. (Internet Era writing license, an emoticon!) I think that is one of the most worthwhile things to do in our era. Comprehend human collectives towards making them happier, more productive, innovative, and powerful.

Doing so we are also creating teams (dynamic structures inside the companies) that create and hold SHARED NEW KNOWLEDGE. Actually our outcome, our product, our result has the characteristics of our method and technology. And thus, it is a corollary of all the logic I stated before in this article.

Beyond Artificial Intelligence is the Hybrid Intelligence. And beyond Hybrid Intelligence is going to be a hybrid kind of living that is beyond our flesh-and-bones human paradigm. We are now creating it within SynerGears, but when we merge deeper with computing systems, it is going to be in the realms of what we now call science-fiction, and will NOT BE!… fictitious anymore. It is going TO BE, simply put: ADVANCED SCIENCE. It will consist of just another way of consciousness, more related to how Neo defeated the machines in Matrix III. Is this worth living for? It is this worth fighting for? I think it is. And that is why I work hard on designing this type of system.