Human level AGI and AI take off/singularity is VERY near two hypothetical scenarios
Posted: Thu Aug 19, 2021 12:51 am
I am convinced that Human level AGI and an AI take-off/singularity is very near. I recently purchased a Ted Talk by Google I was shouting about it on this site so you probably know, if you want to see this Ted talk message me and I will give you the link with login details as I have only given it to one person so far. (I won't be sharing it with too many people to prevent Ted cancelling my purchase so get in fast)
So Google is working on a AI called pathways that can learn from text, audio, video, images (maybe more inputs) and the AI can learn an infinite amount of tasks generalizing its current knowledge to help it learn a new task.
I don't know the process a company like Google goes through when making an AI. They have announced semi publicly that they are actively making it so if anyone knows: How far away they probably are from having the finished product?
Please let me know.
Ok so lets say Google builds this AI called pathways in 1-3 years.
Scenario 1 human baby like AI
An AI that can learn an infinite amount of tasks and learn those new tasks through generalising it's current knowledge turns out to be true AGI with infinite potential and is only limited by hardware. In late 2022 - late 2024 the AI called Pathways is declared finished knowing say 1000 tasks. In the following 2-6 years Google adds new tasks to the AI and the AI eventually knows enough things to learn the ability to learn tasks on its own, the AI eventually becomes as smart as an Adult human by late 2024 - late 2030.
Scenario 2 current trend continues
In late 2022 - late 2024 the AI called Pathways is declared finished there is hype for a year and a half or two like with GPT3 until something better comes along ... AI continues to progress exponentially and hype surrounding large breakthroughs only lasts a couple of years before something better comes out. The writing on the wall for human level AGI was already there with Google pathways and human level AGI is created 4-8 years after Google pathways was finished in late 2026 - late 2032.
So Google is working on a AI called pathways that can learn from text, audio, video, images (maybe more inputs) and the AI can learn an infinite amount of tasks generalizing its current knowledge to help it learn a new task.
I don't know the process a company like Google goes through when making an AI. They have announced semi publicly that they are actively making it so if anyone knows: How far away they probably are from having the finished product?
Please let me know.
Ok so lets say Google builds this AI called pathways in 1-3 years.
Scenario 1 human baby like AI
An AI that can learn an infinite amount of tasks and learn those new tasks through generalising it's current knowledge turns out to be true AGI with infinite potential and is only limited by hardware. In late 2022 - late 2024 the AI called Pathways is declared finished knowing say 1000 tasks. In the following 2-6 years Google adds new tasks to the AI and the AI eventually knows enough things to learn the ability to learn tasks on its own, the AI eventually becomes as smart as an Adult human by late 2024 - late 2030.
Scenario 2 current trend continues
In late 2022 - late 2024 the AI called Pathways is declared finished there is hype for a year and a half or two like with GPT3 until something better comes along ... AI continues to progress exponentially and hype surrounding large breakthroughs only lasts a couple of years before something better comes out. The writing on the wall for human level AGI was already there with Google pathways and human level AGI is created 4-8 years after Google pathways was finished in late 2026 - late 2032.
