Replicating thinking is not thinking.
The attraction of AI, mapping cognitive processes, finding out where 'conciousness' resides and other grand sounding questions lies in a fundamental confusion. That is: the replication of thinking activity so that what you have looks like someone thinking is NOT the same as someone thinking.
Replicating thinking is quite boring- as Alexa, predictive texting, smart algorithms, smart cars and other computer marvels have indicated. Even a computer that makes music better than Beethoven would not be as interesting as Beethoven- it would simply be - a clever machine. The reason is - thinking is the tip of the iceberg of being human, and being human is the real mystery. Making something that replicates thinking is like making a robot that can dance and sing- at first it is WOW and then it is 'can I knock this thing over'...
Fear of AI is misplaced. Even if we make machines that can do all kinds of thinking activities including creative ones we will still have machines which need turning on and off. (Of course you could program a 'purpose' into a machine so it turns itself on and off but that still isn't 'thinking'.) Animatronics can now deliver a creature that looks and behaves very like a cat- but it isn't a cat. Since thinking is something only humans can do (it is a defining human activity) then any machine that immitates thinking is simply doing that- a rather limited thing.
But we get excited and use 'thinking' in the wrong way. What we mean is 'immitates thinking in this way'. That takes the excitement out but also makes things clearer. Of course people are using this confusion to raise money for expensive projects that seem to promise endless life etc. Obvious idiocy like 'the singularity' fall into this category. Even if all your memories could be uploaded how would you choose which memories to 'view' at any one time (which is 'thinking')?