The reply by “nv-vn” here reminds me of how we’ve lost track about the original purpose of AI. We’ve lost track so much that we now stick a ‘G’ it the middle.

A lot of the comments here seem to focus only on the progress we’ve made rather than what AI should be (if it was true to the name). With “AI”, sure I can tell Siri to send a certain email to a certain person. That’s pretty damn impressive when I think back to what was possible only a few years earlier. But it’s not intelligent. Nowhere near. And honestly, it doesn’t save you that much work. It just performs some preprogrammed tasks given some unambiguous input that follows a regular pattern. And even that trips up Siri a ton. Yes, it’s cool that you can identify pictures or have a program read your handwriting, but none of that gets us any closer to a program that behaves like a human when I tell it to do something (instead of sending an email or text that I specify, it could draft one on its own; instead of showing me a Wikipedia article when I ask what something is it should explain to me in layman’s terms and give me a summary). AI is impressive in terms of progress, but is it really impressive in terms of what it promises to bring? Is it really even progressing in the right direction? Personally, I don’t think so.

When working on hard problems, lots of spinoffs and interesting things happen, and it is easy to lose track about what it is we are trying to achieve in the first place. Personally, I’m interested in solving AGI (or, more realistically, trying to). And doing so requires the following: learning to program, learning maths, information theory, probability theory, machine learning… Because talent in any of these domains is scarce, continuing to pursue the original goal means saying no to a lot of opportunities along the way.

That is why I’m not that interested in deep learning: it’s a highly valuable skill and achieves state-of-the-art performance on many tasks, but ultimately, it’s not much more that machine perception. To solve AGI, we will need more than that.