4 ms·
Sold on the premise that when AI gets here it will change everything. What I don't have a good grasp on is how fast it will come. Recent feats of AI are very im
by cbau 6y ago
Sold on the premise that when AI gets here it will change everything. What I don't have a good grasp on is how fast it will come. Recent feats of AI are very impressive, but it's hard for me to put it on a trendline that would line it up with massive changes coming with 10 years. Predictions around AI have made similar claims for the last 50 years. Why is it different this time?
- matthalvorson 6y agoI'd recommend the book Life 3.0, the author surveys a large number of AI researchers to answer this timing question (I think 95% said AGI is guaranteed in the next 50 years iirc), and also discusses why this time is different than the times in the past, like in the 60's when a group of researchers thought they would make significant progress towards AGI over the course of a summer
- jfk13 6y agoI suspect that if someone asked the same questions a decade ago, or two, they'd have gotten similar explanations of why "this time is different". AGI within the next 50 years? I don't think we have any idea, really. We don't even know what "intelligence" means.
- A12-B 6y agoI try to keep up with the industry, just to see where it's going. From what I can tell, AI (in the general sense) is so flexible it can be applied to just about anything with observable data points, which is basically everything. Are these applications useful? I think right now they are impactful, not necessarily useful. EG: we can already copy someone's voice perfectly with just a small amount of audio using AI. But in practise, not much has come from this incredibly remarkable feat. So what does AI need to get to the next level? Not much but time to mature, all the tools are already there.