Can Intelligence be Artificial or can Artificial be Intelligent?
AI (Artificial Intelligence) has been in the general subconscious for eons; Animal Intelligence with Animal God or Geological Intelligence, with rocks or places that have super powers, etc.
More recently, with the creation of Science-Fiction and anticipation, AI has been at the core of humanity’s most vivid dreams/fantasies and fears as a matter of fact.
Technologically speaking, it has been in the works for quite a while too. Neural networks and AI have started to become real sciences (not fictions) with the dawn of IT, but were quickly limited by processing power. It went into “sleep mode” until computers recently became fast enough to make a big step-up in AI. Moore’s law was the fuel of AI.
And also thanks to the amount of available data that came with it.
So it was the rebirth of AI.
AI comes in mainly three flavors:
Algorithms that iterate to solve problems based on human inputs. While this method can process bigger volumes faster that any human could, the results often fall below the quality of what a trained individual could produce. In addition, these algorithms are not capable of evolving on their own.
Learning with help
A system that is capable of finding its own way of solving problems based only on data inputs tagged with the correct associated answers. This method can also process bigger volumes faster that any human could, with results usually as good, if not better, than those of a trained individual.
Learning without help
A system that is capable of finding its own way of solving problems based only on data inputs without the previously mentioned tags. While this method can process bigger volumes faster that any human could and will learn, it is still difficult to ensure what and how. Similar to the learning with help method, results will usually be as good, if not better, than those of a trained individual.
As you may remember, this very last example made headlines such as the following one from the Telegraph: “Microsoft deletes 'teen girl' AI after it became a Hitler-loving sex robot within 24 hours”.
Is it the most worrying part? Or is it that probably a good share of the Humanity, i.e. Human Intelligence, thinks exactly that way?
As René Descartes once said : « L'intelligence, c'est la chose la mieux répartie chez les hommes parce que, quoiqu'il en soit pourvu, il a toujours l'impression d'en avoir assez, vu que c'est avec ça qu'il juge. »
In English : Intelligence is the most fairly distributed thing in the world, for each individual, whatever he’s granted with, thinks he is well-endowed with it since it is with that amount that he actually thinks.
So to that extent, Artificial Intelligence has reached its goal: match (part of) Human Intelligence.
But is it a good or a bad thing for us?
It underlines the well-known IT phrase that says ‘Garbage In Garbage Out’, since bad/poor learning/education will most certainly lead to bad/poor intelligence.
So, just as we need to “watch” mean/ill education for people, we need to “watch” data sets for artificial intelligence.
Intelligence is being able, presented with information, to make the proper/good decision autonomously; and this can now be done by either Human or Artificial Intelligence.
But if the “training/education” is improper, so will be the result.
So AI can be as high or as low as Human Intelligence…. Is it what we dreamt of?