Updated: Feb 5
#PERFORMANCE OF A.I. IS FAR BEYOND WHAT WE SEE IN MEDIA
Forbes recently published a paper explaining why artificial Intelligence without the right data is just ... artificial (Link in the comments).
I found the paper quite outdated because it no longer addresses an important A.I. challenge and somewhat minimises the impressive results obtained with A.I.
For example, I was really impressed by the recent interview of an A.I. being trained in Google labs and would definitely love having a friend with such knowledge, conversational intelligence and emotional potential (see Lamda interview via Google search).
Unfortunately this is not yet possible… Not because of performance, but probably because the level of social interaction I expect would require far too much computing power and energy to make it available (at a reasonable price).
I am more inclined to believe the major roadblock that will prevent A.I. from reaching its full potential is energy requirements. Artificial intelligence systems require vast amounts of energy to be able to learn and deliver results that are comparable to those of humans.
Research efforts should now focus on developing Artificial Intelligence systems that can operate efficiently with minimal energy and uncertain data, enabling them to deliver a level of language, emotion, understanding, calculation and decisions similar to what people can do.
I guess that this approach of Artificial Intelligence will truly have good chances to take off and become applicable in a vast number of areas.