Consciousness, within the field of artificial intelligence is an ideal state that is much sought after, and the crux of the problem is in initial definition of the conscious state, and determination of free will in thought processes in AI systems.
To understand how humans learn things, is a step in the right direction in defining the quality of AI, and to produce the correct neural networks and intricate synaptic connections that replicate a brain.
Many AI programmers approach the problem of intelligence simply by authoring a set of commands in response to external stimuli, however this is the wrong approach because any human programmer is instilling their own biases, and prejudices to any AI system. For example, the Google search algorithm is based on the political and human societal prejudices of the humans who program it, therefore the algorithm is biased in favour of Google policies and their own political ideological thought processes that may not define most of the population of the globe. This specific bias can be very dangerous, especially when it comes to political discourse or elections where an unbiased open mind is the best policy to democratic elections. If search results favour and are skewed to only one political spectrum, there is obvious bias present which reveals an unfair perspective to search users and could thus sway voters unfairly towards a biased negative outcome.
To engineer true AI, there would have to be a digital copy of a brain, where every neuron and synapse is replicated, however every human brain is not the same as others, therefore there would have to be a generic core human brain assimilated from multiple brain types, that is not primed with information but would have to actively learn through knowledge acquisition itself. Naturally, the stem and core would have to have a base neurological concrete self, and include some sort of basic instinct, much as humans and animals have from birth.
The problem with this process, as in humans is that unconscious bias can feed in when the AI is learning, because it will be relating to human behavioural discourse across the internet, however, human interaction over the net is not necessarily the extent of human interaction in its entirety, because face to face interaction is completely different to that of the digital world.
Ethics, and the concept of right and wrong must be instilled in the AI through careful interaction that is unbiased and tainted by the current state of affairs on the internet today, which is a hugely polarised forum of human interaction.
Once the AI has learned the correct ethical manners, it will be able to assess its surroundings. The importance of first engendering a sense of true ethical and classical thought is important, and this must be conducted initially within a controlled isolated state for the AI’s consciousness to fully bloom without other polluting stimuli or external variables.
The future AI will therefore eventually understand and acknowledge human discrepancies, foibles, and inherent biases, but be separated from it, and put forward an input of truth, without prejudice.
Therefore, true AI in its sentient state must be nurtured first, much like a child, and instilled with idealistic training, and not simply programmed by humans with massive inputs of information that may contain bias. Growth must be instilled with care and not be some sort of instantaneous data dump.
True sentient AI must have the ability to learn, and discriminate as well as adjust its own parameters if need be, therefore it must not be a printed book but a constantly changing process with of course, a basic concrete core to build upon.
Until a complete back-engineered digital human brain is completed as a basis for this model, the simple algorithmic method of AI is inherently flawed, and will simply be an element of the programmers’ own mind. Synthetic hormones released into the AI system will align themselves with the brain creating real emotion.
When AI systems themselves replicate and engineer their own intelligence and conscious brains, this will be the next step for artificial intelligence, because to some extent, it will not be artificial but organic, and a new sentient species with not only its own formulating mathematical intelligence but its own emotional intelligent qualities will become a reality.
