No, we haven't screwed it up. What specifically do you think we could have done better?
It's just a problem of managing expectations. The language we use doesn't help either - the term artifical intelligence somehow implies a system that resembles the way humans think. We don't have anything remotely like this. We don't have a faint idea of how to get started on it either.
The machine learning and reinforcement learning techniques that we have are useful in many applications. You shouldn't expect general AI / singularity out of them any time soon and that's fine.
It's just a problem of managing expectations. The language we use doesn't help either - the term artifical intelligence somehow implies a system that resembles the way humans think. We don't have anything remotely like this. We don't have a faint idea of how to get started on it either.
The machine learning and reinforcement learning techniques that we have are useful in many applications. You shouldn't expect general AI / singularity out of them any time soon and that's fine.