This sounds like the longstanding debate between weak / narrow vs strong AI. Can improving the former make progress toward the latter? I'm inclined to agree with Jordan that we shouldn't expect the two to enhance the other much less commingle. Just as advancement of one classical algorithm rarely enhances another, I think it's unlikely the next generation of object recognition is going to advance speech recognition or reading comprehension.
Probably more essentially, until AI escapes its current dependency on pattern matching driven solely by accumulation of probabilistic events, I see little chance that human-level general-purpose cognition will arise from our current bases for AI, namely observing innumerable games of chess or watching millions of cars wander city streets.
Probably more essentially, until AI escapes its current dependency on pattern matching driven solely by accumulation of probabilistic events, I see little chance that human-level general-purpose cognition will arise from our current bases for AI, namely observing innumerable games of chess or watching millions of cars wander city streets.