Whatever the complexity, the fact is that humans can play go at a level that currently no machine can. And likewise for face recognition, language understanding, translation, etc.

I guess our disagreement stems mainly from the definition of "solve". I sympathize your insistence on solving every instance. But we know that's a dead-end, and also not how humans do things. Instead, I'd settle for a definition that merely asks for intelligent behaviors, which may falter from time to time (as humans do). For pragmatic matters that's all it takes, i.e., win sufficient number of games in chess, drive through the dessert or city blocks automatically, understand natural-language text at the level similar to human.

Conversely, this will help research on brains. If we build an information-processing model that simulates human behavior, it may shed light to how brains work. Cognitive science follows exactly this path.