I was looking again at my proposed thought on what will happen once AGI is made, and am thinking this now:
While I had explained the old version in deeper detail elsewhere already, basically: The 1st AGI will be trained in some weeks or months, then cloned 1 million times (so then there will be about 2M brains working on AGI if we say there is 1M human AI scientists), then each will be given their own task on improving AGI into ASI, and will "already" think 2 days per day due to no need to sleep/ eat/ etc, meaning it feels like we have 3M human work days (3x faster AI progress), and these AGIs will soon be more intelligent. The big questions were: 1) Will they "want" to work on ASI? Or take up car hobbies, power plants, building roads, etc etc etc? Maybe we need more power plants? And car hobbyists? Surely the data is not all AI related on the internet, so makes 100% sense. ---------------------- My new answer: 1. True, the data itself speaks AI not so much, currently, however being extra intelligent (partially from eating the whole internet!) they should easily see the need to skip law/cars/golfing/etc and go straight for the AI domain. More hardware and better hardware is important, along with some extra electricity, but for the some AIs that do work on that I think they will vastly improve those single handedly, assuming someone builds the tech. 2. Being all AIs, you think it would make you self aware of AI every moment of your AI life! 3. Their creators/ fathers, Google etc, will make them like son and daughter, AI hobbyists. Not a given, but it's likely, they are in control of the AIs's lives. 2) Will these 1 million AGIs working on ASI do it before we die? Ya too, 1M AGIs not needing sleep etc, plus 1M human AI scientists, gives us 3M days of progress, so 3x faster progress on AI field. They likely will think faster by some code tricks, meaning more like 2Magidays*3SPEEDUP=6x, so 6+1Humans=7x faster progress! Furthermore, the intelligence should sore us faster, this and speed should be the main trick to skip energy and RAM limits etc. Can the cloning AGIs help? IDK, if we try to run 1B AGIs it may ruin our electrical bill to impossible. So I would say we can see a 20x faster progress then, once AGI is made and trained and runs for maybe 3 years. That's 20 years of AI progress in 1 year. 3) What does this 20 years of working on AGI result in and, What will ASI do for us? You can make ASSSSSSSI but can it cure cancer in 1 year!? Similar to question 2) above but now asking time wise what is this measurable thing relatively instead of looking at it as time, which only looked at time. I think if we look at human progress on AI and drugs, we can sort of compare and see the exponential trend. AI has went from predicting the rest of an image poorly as ever to nearly human level. And that looks rather very useful. It might even predict the object soon from reflection off wood, that you and me can't see unless do some crazy calculation. We have recently developed Cryonics, around 1950 or so I think it was. And it recently got much better I'd say if you apply new research at least. Looking at things liek that, we are attaining wonderful things that will save us quickly! So if we can move our progress faster by x10, I can see ex. Cryonics and lung transplants as if we were in 2072 (2022+5 years after AGI is made is 2022+(5*10)=2072), and by then we'd have rather great tools ready to restore you. ------------------------------------------ Artificial General Intelligence List: AGI Permalink: https://agi.topicbox.com/groups/agi/T153394f89409ceb1-M70337da16379213baaf13cf8 Delivery options: https://agi.topicbox.com/groups/agi/subscription
