We’re continually raising the bar, but surely that has to stop someday?
If you showed one of our modern personal assistants to someone from the nineties, they’d be amazed by our advanced AI. But in reality, we’re “just” doing some voice recognition and identifying some meanings in that to perform one of many pre-defined actions. Yet, a layman from the past would easily file it under AI, whereas we are more likely to categorize such functionality under regular old software things.
A stronger example may be the applications we use neural networks for. Broadly speaking, we’re simulating the mechanics of the human brain and applying them to a given task. This means it can learn to perform complex tasks in even more complex ways. And yet here we are, noting that it is really cool and useful tech, but dismissing it as a form of intelligence.
Maybe something will only be labeled AI once we have no comprehension of it anymore whatsoever? Once it can develop itself completely out of our control, learn and apply without any guidance whatsoever? “General purpose AI” may be the only thing we’ll ever consider true AI, and I’m not sure that’s completely justified.
Regardless of labels though, we’ve made cool things, and we’re working towards even cooler ones!