This paper starts with an explanation of how deep learning works: as a series of units each with elements whose output is the weighted sum of inputs from previous units.  The paper asserts that this is probably not the way the brain works and also we cannot follow its logic through the intermediate units to justify its answers.  The paper claims therefore that AI “thinks” in a very different way from people and it is misleading to anthropomorphise AI by using language such as this.  For example, we discussed whether AI has "knowledge”.  It was suggested it might be better to think of AI as a very good abductive reasoner - giving the most probable answer from the statistics of its trained memory.  Machines and people are different and neither should be thought of in the language of the other, but we do use technology even as an extension of our selves, rather as a blind person uses a white stick.  Technology, people and culture develop together.  It is argued we need to use and apply the humanities to envisage, predict and guide how AI, humanity and our culture may and should progress.  It needs to be to benefit people, hopefully not to harm them or to make more money for the rich and powerful who “own” the technology.