Home / Blog / The evolution of Intelligence

The evolution of Intelligence

Scientists have observed that the increase of intelligence, for what can be assessed about our ancestors intelligence, is linked to the amount of blood reaching the brain (that can be estimated by looking at the opening in the skull to let blood vessels into the brain). An interesting observation that might not be applicable in 2 million years time when our descendants will try to gauge our intelligence, since our intelligence may evolve through a symbioses with our artefacts. Image credit: University of Adelaide

To talk about the evolution of intelligence one should first “define” what intelligence is and based on that definition provide a ranking of intelligence. Once we have that we could start discussing intelligence evolution.

Scientists and psychologists are still struggling on this so in this short piece I would take the intuitive definition of intelligence, the ability to interact smartly with the environment, also using conceptual models. In this sense we can distinguish insects (that are surely smart in their interaction with the world) from a number of animals that seem to have the capability of thinking about the possible interaction before engaging in a specific one.

Being more intelligent would mean to be able to create more sophisticated representations of the world and being able to assess the variety of options selecting the one fitting best. The bigger the areas that can be addressed and the better way they are addressed the more the intelligence.

A machine can have a very good “skill” in performing an interaction but in order to be intelligent it would need to create a representation of the context in which that interaction takes place.

This is now a possibility. Artificial Intelligence is commonly used in a variety of areas by machines as different as a digital camera, a robot, a virtual assistant.

What we have seen in these last years is a tremendous growth (thanks basically to increase processing capability and larger data sets) of the smartness of machine and of the areas where they can be smart.

Actually, machines can be smarter than us in several areas making intelligent decisions that are better than ours because they can take into account more data in a limited amount of time. AlphaGo has exceeded the intelligence of the Go world champion in playing Go (and coming up with unexpected moves that indeed followed from an internal representation that was broader than the human’s one).

Although we have several examples of machines that are more intelligent than humans is specific areas we do not have, yet, a machine that is more intelligent than us in all areas. This is what is known as AGI: Artificial General Intelligence. If a machine would be able to be intelligent in any area where a human can interact with the world (including with other humans of course) than that machine would have “our” intelligence. So that is a sort of evolution we can have: from today’s AI to tomorrow AGI. Between the two there are so many hues that it is actually difficult to determine if a machine has reached the AGI point. What we are going to see is that machines will be applied to more and more areas, eventually reaching a point where they could substitute us, in terms of intelligence.

A further step is considered to be ASI: Artificial Super Intelligence, a point where machines would be able to exceed our intelligence, both in areas of applications and in performance (making better decisions).

As an example, that sort of machine would be able to come up with a unified theory of gravitation and quantum physics, AND, it would be able to write a sonnet in its spare time and make some jokes about it with her (no more it!) human colleague.

One of the problem, of course, would be if we could appreciate that machines is more intelligent than us, since to do that we would need to step out of our box and look both us and them. Is a dog (that for sure is intelligent) appreciate that we have a Super Canine Intelligence? I doubt so!

Another aspect is that if a machine can indeed reach the AGI stage, well, at that point it will also achieve the ASI stage since we recognize that toady’s machines are better than us in a few areas. Once they will be as good as us in all areas, they will of course remain better than us in those few areas, hence they will be, overall, better than us, q.e.d.

How do machines get more intelligent? Well we have finely tuned ways to make them intelligent (in few constrained areas) and one of them is through machine learning. Now, I am mentioning this one in particular because in these last few years machines have become capable of self learning, both by looking at their environment (including other machines and humans) and by cloning their “minds” to challenge themselves in finding new models and solutions. Notice that the word “model” is crucial. They are not just becoming smarter in making decisions they also change their “mind”, they grow their intelligence and because of that they can make better decisions. Also notice that I have being saying “making decisions” not “taking decisions”. When discussing intelligence decision making is more important than decision taking, this latter can be a sort of mechanical progress evaluating the (many) possible outcome to take the better decision, whilst the making of a decision requires intelligent capabilities.

There is another “twist” to consider. Once machines are interacting with their environment, the result is an environment that displays some sort of intelligence, the one of the machine. If we have several machines interacting one another the overall intelligence has components that are tied to individual intelligence but it can also show an intelligence that is emerging out of those interactions. This is the situation of swarms of insects, flock of birds… The ensemble is more intelligent than each of its individual component.

The emergence of intelligence is a typical characteristics of symbiotic complex systems (systems that cannot be simplified without losing their characteristics). This is something that is being studied in the IEEE Symbiotic Autonomous System Initiative and it is being studied in relation to the symbioses of machines and of humans with machines.

The emergence of a symbiotic intelligence among humans and machines can be an answer to the concern that eventually machines will become more intelligent than humans, hence leading to the end of our species (or as somebody predict to become slaves of machines). Indeed, if we buy into the idea of symbioses we could say that as machines get smarter, we get smarter in a never ending story. The crucial point is to join the machines in this growth.

About Roberto Saracco

Roberto Saracco fell in love with technology and its implications long time ago. His background is in math and computer science. Until April 2017 he led the EIT Digital Italian Node and then was head of the Industrial Doctoral School of EIT Digital up to September 2018. Previously, up to December 2011 he was the Director of the Telecom Italia Future Centre in Venice, looking at the interplay of technology evolution, economics and society. At the turn of the century he led a World Bank-Infodev project to stimulate entrepreneurship in Latin America. He is a senior member of IEEE where he leads the New Initiative Committee and co-chairs the Digital Reality Initiative. He is a member of the IEEE in 2050 Ad Hoc Committee. He teaches a Master course on Technology Forecasting and Market impact at the University of Trento. He has published over 100 papers in journals and magazines and 14 books.