We humans tend to look for patterns in text, and tend to interpret them based on their experiences with other humans. Years ago I wrote a stupid simple program

We humans tend to look for patterns in text, and tend to interpret them based on their experiences with other humans. Years ago I wrote a stupid simple program that responded to typed in remarks. Each typed remark would be added to a cache of remarks. The program then parsed the new remark into words. Then it looked through the cache to find an old remark with several strings in common with the new one (else a random one), and replied with it. (It also marked the old and new remarks as 'used' so it didn't repeat itself in a session.) So, any apparent intelligence was human. When people typed stuff in, they tended to do their best to interpret the reply as intelligent. They might puzzle until they thought they 'understood'. Several people got emotional; some of the regurgitated remarks were not so nice. They might then add their irritation to the cache! One day, with several people watching, it took quite a while but found no good replies to a question, and replied: "I like big tits." That was the end of the demos.