Should we limit the AI development?

This may sound technophobic, but should we limit the AI development to the point it remains just a helpful tool and nothing else? I'm actually a bit scared of the possibilities AI may reach one day as I see what it can do now (it's not AI at its fullest, I know). Humanity has a tendency to break the limits even when it understands that limits are set for their safety. I just don't want to see Skynet rise :D