The problem is not "computers," it's intelligence itself. We still don't know how even the simplest neurons actually work, nor the simplest brains. And we're barely any closer to scientific definitions of "intelligence," "consciousness," etc than we were in the 1800s. There are many decades of experiments left to do, regardless of how fancy computers might be. I suspect it will take centuries before we make dog-level AI because it will take centuries to understand how dogs are able to reason.