Co-evolving with the future of AI
In some sense, from a third-person point of view, human brains and computers are very alike, they have inputs, outputs and the main thing happens in between the two, they are like black boxes. Many of the processes running in these black boxes are profoundly subsymbolic, composed by myriads of electrical signals dynamically sparking through a vast and complex network of neurons. Trying to explain them through human language is almost impossible.
“The most important part of an AI is the data you use to feed the system”.
There is a fundamental problem in the concept of AI. If you ask someone where the intelligence is located in the “input -> black box -> output” scheme they always tend to put it in the middle of the process, but the truth is that the most important part of an AI is the data you use to feed the system. It doesn’t matter if you have the best algorithm ever, if you feed it with low quality data, the results will be poor and mediocre (Garbage In, Garbage Out).
So the first step to create great AI systems is to feed them with good quality data. Next, we have to think about the following: out of the wide range of problems we want machines to solve automatically, is there any subset that require self-consciousness to be solved? Do we really need to cope with all the human psyche complexity in order to achieve human-like intelligence in machines?
Elon Musk claims that if we want to prevent the tyranny of Artificial Intelligence, humans and machines must develop a symbiotic relationship. We need a brain-computer interface that would enable humans and super-intelligent machines to coexist peacefully. And that could be achieved “by improving the neural link between your cortex and your digital extension of yourself”.
This is the road AI is taking. Little by little we are adding layers to human capacities: taming, new technologies… We can also extend sensory abilities and physical abilities (exoskeletons) but, can we extend mental abilities? To extend a mind we need appropriate interfaces. The scheme is as follows:
Both, artificial systems and humans, have input/output subsystems. In traditional human-machine interaction, data is interchanged via these subsystems. But we could eliminate “unnecessary” proxies, thus removing the boundaries between the human body and its extended capabilities. We will directly connect mind and machine, instead of seeing the information in a screen, we would feel it directly, as an extended mental capacity.
Therefore, the main issue is not determining whether or not machines will be more intelligent than humans, the real problem is how we will co-evolve with machines. The decisions we have to make concerning an ethical code for AI must be more focused in this area.