Why do you think the timeline for AGI is what it is?

This may be a dumb question, but I'm curious to hear thoughts on why people think the timeline for AGI is what it is. <br><br>The longer I spend reading this subreddit and other resources on AGI, the more I'm convinced that it's going to be later rather than sooner. When I say later, I mean 2075 or later is my guess.<br><br>I think the thing that's really convinced me that it's later is how difficult it is to build narrow AI. Like, the people who are building it are like freaking geniuses. Theyre not just smart. They're like off the charts smart. It takes years and years of research to develop what amounts to a narrow AI that can do anything practical. And that's just for something like AlphaFold or Perplexity. AlphaFold is narrow AI! Like, the amount of research that went into it is nuts.<br><br>So, when I hear people talking about how AGI is going to happen in 2028 or 2035 or whatever, I'm like, yeah no way. I just don't think it's realistic. It's like, we're so far off from something like Perplexity which basically just does what AlphaGo does but for words. And that's just for words! Imagine going from a narrow AI that can just do words to something that can do everything. Like, I just don't see it happening for a while.<br><br>I think the major things that need to happen will be, evolutionary algorithms and the ability to program AGI via RLHF. Until we get to an AI that can do a ton of tasks, I think we're really just in the narrow AI phase, and when I think about what needs to happen for us to get to that point, I just don't see it happening for a long time. I think it's going to be a long time before we have something that is close to AGI.<br><br>I think we'll be lucky if AGI happens in the next ~40 years.