When we think about Artificial Intelligence and Robots, we naturally assume the Robots to assimilate humanoid characteristics. A head that sits on a neck connected to shoulders with two arms and two legs. With a brain that thinks and behaves similar to humans.
But perhaps we should not view Robots in this fashion. Maybe we should disassociate the human form with that of the robot. They may have wheels instead of legs. They may have multiple appendages instead of two arms. They could have eight like an octopus. And why only two eyes, why not thousands like the bumble bee. And why not have x-ray vision like the bats.
We would not want to limit the body type to mirror humans, as we were formed to survive based on current environmental conditions and have evolved over time. Robots don't need to evolve, they can be built precisely from the beginning.
We currently have Intelligent devices in society today. Have you ever stopped for a red light, waited a while, then drive when it turns green. Is that not a form of Intelligent behavior? What about IVR phone systems, they can interpret key pad and voice recognition and route your call accordingly, it can provide information and even perform services. Is that not a form of Intelligence.
Why must we associate Artificial Intelligence with Robots which mimic human form? You can have AI without a body at all, or a modified version. Do we not already have derivatives of AI in action today? I think we do.
And if you think about computers, with Neural Networks, which can be trained to understand past events and associate future behavior by applying weights to neurons and connecting them to other neurons. It simulates an actual brain but lacks a certain 'Intuition' at this point in time, it uses pure logic. Yet humans can do quick associates in the minds to derive conclusions, without knowing the actual steps they took to get there, a mixture of seeing pictures of events in their mind and recalling memories to produce a result.
So Neural Networks are a form of AI, so are pattern recognition systems, and predictive applications and having programs learn on the fly with no input. They recently had a computer learn a few Atari games, on their own, by figuring out the rules, achieving better results than humans in some instances.
What would it take to have an AI system pass the Turing Test? Ability to reason, process information, insinuate answers based on analogies, etc. Doesn't say anything about physical form, just a way to receive questions in the form of input devices, a way to process that information to respond with the correct results in the form of output.
I don't necessarily think we need to mimic the human brain. I think we need alternative ways to achieve the same results. And that would be based on the current technology. Of Big Data, Neural Networks, Deep Learning, fast Processors. But in order to scale, it would need to have multiple Neural Networks tied together through a network. Each Neural Network specializing in their domain knowledge expertise and then linking them all together. So you could have one that knows plants. One that knows Trees. Another that knows Mammals. Birds. Reptiles. Stock Markets. Politics. Religions. Cars. Boats. Etc. Link them all together for a conglomerated humongous Neural Network. Constantly fed new information. An All-Knowing Encyclopedia Brain like device. You could call it Hal 9000 from Space Odyssey but that that one didn't go so well. It wouldn't necessarily have a body, but it could. You could connect Robots to the Brain like structure so they all have access to the same repository, in real time. Kind of like the Borg, constantly assimilating information.
I think the definition of Artificial Intelligence should be re-clarified. I think aiming to mimic the human brain is too low a standard, we should reach for something higher, and in order to do that we need to take alternative approaches. We may never figure out how the brain works, how nature works, so in the meantime, let's figure out a way to get true AI by any means possible.
And here's a follow up post ... http://www.bloomconsultingbi.com/2015/02/ready-or-not-automation-and-robotics-on.html
I signed up for the Hortonworks Certified Associate exam last Thursday. Figured if I sign up, I'd have to take the test. And if I tak...
Saw a post today on Twitter, " Microsoft releases CNTK, its open source deep learning toolkit, on GitHub " This is big news. Be...
This blog post is in no way an attempt to steal other people's work. It's basically an conglomeration of notes from research I did...