The first two are logical problems that can be approached systematically, because go and chess are idealized systems with simple sets of rules. They're about as

The first two are logical problems that can be approached systematically, because go and chess are idealized systems with simple sets of rules. They're about as relevant to understanding subjective and culturally dependent, ultimately irrational notions as "They said computers would never draw a million triangles per frame" To say that Siri understands natural language is just laughable. It can analyze natural language in a way that is extremely error prone and contextually insensitive, just barely well enough for it to parse Zork-like commands and google stuff for you, set alarms or call people in your contact list. What's at all slightly impressive about it is its voice recognition, and even that is very error prone.