I don't understand what bearing this should have on the probability of AGI succeeding. What we (should) care about is not whether the system is conscious or even human-like. What matters is that it performs well on the tasks it is assigned. I consider it highly likely that we can do better than evolution did, in less time than evolution took.