Hmm....Let me think. I know of 4 ways that come to mind how to test for AGI:
Prediction Score, such as Perplexity or Lossless Compression. If it sounds and draws and sees etc like us. How much money it makes. How many clones it makes and how much it can repair itself. The 1st way is more comparison between AIs, and not human brains and not human level. There is no score there that says how close we are, rather only how much better than last AI it is. The last way is too long for us to test on current brains. The middle 2 ways is all about how much stuffz it does, a subjective way we view it. But general purpose AI (AGI) is what makes it more testable and definable and track-able, (AI) is narrow intelligence. True, since 1950, it has been 70 years of AI progress technically. That's an incredibly long time. During that time, doubling has been occurring of intelligence, from bacteria level to fruit fly to mouse to cat to child level intelligence. Not so much in how much they know, nor how intelligent they are, but all such things, as you can see older AIs lack many cognitive abilities and also data size. Yes some old AIs had a lot of data but Generally, Overall this holds I think. And to get where we want, we need ANOTHER say million increase which would give us from human to human * 1,000,000 in ability. But that isn't going to take another 70 years, it only would if our technology and wisdom didn't improve. But it is, and fast. ------------------------------------------ Artificial General Intelligence List: AGI Permalink: https://agi.topicbox.com/groups/agi/T91a503beb3f94ef7-Mb59a2781a1865ff36136c603 Delivery options: https://agi.topicbox.com/groups/agi/subscription
