I don't know what I had eaten or drunk a few nights ago but whatever it was, it spawned a vivid dream concerning artificial intelligence and in particular referenced human replica robots.

The dream touched on the concept that a human AI replica had developed a conscience that gave rise to a deep issue within the AI brain around the characteristics of a person.  In my dream, the AI "robot" was concerned about the impact of itself looking exactly like another factory built AI human replica.  The AI brain had learned that every single human is unique with its own unique characteristics - even identical twins have some slight variations or differences (slightly different tone of voice, slightly different hair style, placing of mles of freckles etc.) and it was understandably concerned that it would look, act and sound exactly the same as any other factory manufactured human AI replica.  It would in fact, lose or indeed have no unique identity (apart from, perhaps, a different model or serial number).

After I awoke, I laid in bed for a few moments to try to fully contemplate the content of my dream - of course I could not fully recall every last detail of the dream, but the overall concept stuck with me quite clearly.  It led me to the conclusion that true AI will never be possible.

Humans have always been captivated by the possibilty of AI robots and we continue to pour huge sums of money into the whole AI and automation concept to try to make these AI bodies more self-aware and able to learn.  To try to make them capable of being human but better in the sense that they don't have the ability to consciously disobey orders or rules and that they are pre-pogrammed to do everything we need them to do perfectly, time and time again.  By this very definition this is not true AI.  To be true AI, the AI would need to have the capacity for free thought, it would need to be able to record experiences and then draw on the knowledge from those experiences and be able to formulate a theory, opinion or some kind of conclusion in order to perform better the next time a similar situation is encountered.

Creating true AI would result in an artificial brain that would be continually evaluating everything it encounters and experiences and looking for ways to improve itself and those experiences - this is good, this is a truly beneficial outcome of developing AI and genuinely what the concept is about.  But, that same AI brain would also be constantly questioning everything it encounters and experiences in an effort to further understand what it experienced, why it experienced it and whether it experienced something in the same way as another being (human or artificial).  This is the less desirable part of AI, the most difficult to control and the thing that will be the failing of AI (whether that be that AI fails to succeed, or that it succeeds just a little too well and potentially becomes a threat to human existence rather than an intended enhancement) and is possibly equally a huge failing in humans.

Additionally, pushing that aside, there is the biological element.  I doubt that AI human replicas are going to be designed with fully functioning internal organs and glands creating and secreting various bodily fluids or indeed fully functioning reproductive organs.
So, if we have the AI brain functioning like a human brain with all of the knowledge of the human body, surely the AI brain is eventually going to develop maternal or paternal instincts and desires?  Can we really predict what an AI brain would do when it realises it cannot give birth to children and experience the maternal or paternal love and emotion that comes with children.  Indeed how would an AI brain cope with love and sexual desire when it cannot actually experience these things?  If the AI brain will be capable of emotion, it will be capable of rage and jealousy.  This is not, I suspect, the desired outcome of those who are developing AI and automation.

Meaning I stick with my conclusion; true AI can never be possible or achieved and indeed in light of the above it's probably good that we are unlikely to ever able to create a true AI "lifeform" - the results could be catastrophic for us all.