re: A Short History of Artificial Intelligence VIEW POST


Interesting read. I would like to add a few points:

  • Artificial means "made by humans/not occuring naturally". It's not fake, it exists, it's real. :)

  • One of the big thing with AI, as you mentioned, is that we're not sure what is "intelligence" exactly. And with every AI breaktrhough, the definition of intelligence shrinks, because if a machine can do it, then it's not really intelligent. Because of computers, we knew doing arithmetics doesn't need intelligence. But chess, chess need planification, strategy, mind game... Of course winning at chess recquires true human intelligence. Then Deep Blue happens, and chess becomes a difficult game but not that intelligent, just a game with a huge combinatorial state space. No, true intelligence must happen when there's hidden information, when intuition is needed, it's being able to adapt and perform well at different tasks... Until a machine does better than us at this. It's endless! And I think it's partly due to people thinking "artificial" means "kind of fake", this way it can never be truly intelligent.

  • About AGI and superintelligence, if we ever build a machine capable of one of these, it will take years (at least!) to be certain that we've actually achieved this. There are still debates about intelligence in animals, and they're closer to us than machines. How are we going to be sure we're not confusing intelligence with fast computation + clever programming + antropomorphic bias?


Thank you for reading! Great points Jérémy!

I agree with you in the point, that the definition of Intelligence and therefore also of Artificial Intelligence is really difficult and it is still being evaluating by experts. It seems like a steady process.
I think the problem here is, that there are many branches, medicine, biology, philosophy and also computer science who all are trying to find a definition for intelligence. Intelligence is not only part of one branch, it is part of many branches. Therefore it is very difficult to find a definition which makes everybody happy.

About AGI and superintelligence:
If I would have to say when there would be AGI and superintelligence I would say at least 40 years. But I actually don't know ;)
Again, I think Nick Bostrom's book 'Superintelligence - Paths, Dangers, Strategies' is one if not the greatest book about this topic. I encourage everybody who are interested in this topic to read it!

About the definition of Artificial:
I still think that Artificial is (kind of) fake. Of course it exists, but it is not real, because then it would be just a clone of something. And if we are building something artificial, we are not just cloning something. Real grass would be grass that would grow, but artificial grass is fake grass, which intention is to substitute real grass to improve it behavior.

code of conduct - report abuse