As for general AI being "promised" in the 1960s - I think that's being a bit harsh.
Real general AI has been a bit like fusion power for the last 40 years or so - it always seems to be 20 years away or so, but actually I think that in both cases real, if slow, progress is being made and we will eventually get there. Taking a "show me the code" approach to something as complex and understanding what intelligence really is, and then trying to capture that in a software system is perhaps underestimating the complexity of the underlying problems just a bit.
