Many definition from Internet
Here is a definition of Artificial Intelligence (AI) that I got from the Internet: "Artificial Intelligence is the set of theories and techniques used to make machines capable of simulating intelligence". And there are lots of other different definitions ... So I'm going to be interested in this one.
I find this definition too general and not everything to be fair. As far as accuracy is concerned, the day when a real general artificial intelligence is written, this machine will not be just a "set of theories and techniques". She will have and will deserve the status of being alive. She will feel demeaned for being just "theories and techniques." These 2 words "Artificial Intelligence" are catch-all today. It must be readjusted using words of fair weight.
Definition too general
Concerning the overly generalist definition problem, the concern comes from the democratization of neural networks. There are more and more people working with these tools, there is undeniably a fashion effect. We tend to forget the symbolic side of intelligence. Working with neural networks that can only reproduce (for the moment) algorithms without explicability possibilities, this amounts to reproducing systems that function like black boxes. It works, it's super efficient, but you can't explain the decisions. Doesn't that remind you of something? Instinct, reflex. The two cannot be explained. Instinct and reflex are very powerful! So I find that in this definition, with the lack of precision and our time which swears that by neural networks, the words "simulate intelligence" seem to me too general and should be more explicit. We must give weight to the two main currents that have always existed, the symbolic side and the connectionism side of Intelligence. We have to readjust it fairly. Reasoning is not generally a reflex. However, automatic reasoning is a form of intelligence. It must be emphasized. Reflection against Reflex. Symbolism versus Connectionism.
One thing also missing for me in this definition is the objective. The goal. The race behind it. It has always been a race, the search for the grail in IT. Super AI. General artificial intelligence. Artificial life. We also have to adjust that.
Why a good definition?
These two words that are "Artificial Intelligence" cannot continue to be a catch-all. Everyone is getting lost, it's time to get over it. I therefore propose this definition which gives both a goal, a differentiation between algorithm and search for the grail, of correctness in the chosen words, and finally of importance to the two main visions that is connectionism (reflex) and symbolism (reasoning). The words reflexes / instincts / detector are debatable, but I couldn't find a better word to make you understand.
So here is my definition:
«The aim of Artificial Intelligence (AI) is to design, at low level, algorithms capable of reproducing, using techniques and theories, certain human tasks in its reflex / instinct / detector activities. At high level, to completely reproduce a living being in its activities of reasoning at least, the singularity will operate for the next.»
So understand that these technologies will be very important and will lead to a lot of change in our habits, our ways of looking at things, our other relationships. I see there an incredible chance to raise us in the respect of others, with machines...