Currently I share that AI lacks enough to stop being a weak type and become a "strong AI" in which the neural network is so developed, both in its ability to learn and in its freedom to decide , that we may have to ask whether we should grant some legal rights to this "existence".
Although it should also be noted that I do not think that it is good everywhere that AI learn freely from their interactions with the world and develop something similar to "porpian opinions", an interesting case is that of the Chinese chatbots "Babyq" and " XiaoBing ", who expressed things contrary to the Chinese Party and were" Purged " (gentle way of saying that they were eliminated), I clarify that we may never know if in reality those things they said in favor of democracy and against the Communist Party were in actually learned from their interaction with users of the QQ network or were a hidden part left by their programmers.
But, not only governments have had to delete chatbots that have become problematic, we must remember the case of Microsoft in 2016 with its chatbot Tay to Twitter, which was expressed in a politically extremist and racist way ... well, no way , they deleted it too.
To close, I think that in the long run the chances of an AI questioning and fully developing the "desire" to know and understand are not zero ... although using the term "desire" on a machine is incorrect ... I think.