🧵 Untitled Thread
Anonymous at Tue, 18 Jun 2024 04:37:24 UTC No. 16240555
I am VERY confident, I have come up with a method that will allow us to achieve human level AGI within 3 years, with a budget requirement below $200mil. I had the methodology reviewed by one person so far, and they are an AI researcher working with a University in California, no names at the moment. They had come up with the same idea themselves, independently of me, but had never bothered to crunch the actual number requirements for the resources because something specific "seemed" insurmountable, but the major limiting factor, is no longer an issue. I performed the math myself for the budget estimate, based off of the real world resource needs. If I truly believe I am correct, do I try to file a patent, or do I try to get a job at an AI company, so I can actually do it myself?
Anonymous at Tue, 18 Jun 2024 05:14:57 UTC No. 16240577
>>16240555
Seems almost impossible to patent the broadest utilization, so I am thinking of patenting one key detail that would be required by anyone trying to use the method.
Input?
Anonymous at Tue, 18 Jun 2024 06:40:58 UTC No. 16240653
>>16240555
Tell me you don't know anything about AI without telling me you don't know anything about AI the post
Anonymous at Tue, 18 Jun 2024 09:04:02 UTC No. 16240748
>>16240653
I know enough about AI to have performed a cost analysis based on true, real world asset costs.
As in "I know the necessary design of the architecture, and built up the estimates off of the exact individual component hardware cost, in totality, required for the specific load."
My approach is not crazy. It is novel. It will work.
It's also not like this is an approach that hasn't already been thought of, in some way.
But I was referring to a specific technology breakthrough that only happened this year, which allows this method to finally work. (The "seemed" insurmountable factor)
I'm not a genius and don't claim to be, which is why I am pretty sure someone already has a patent for the method under the USPTO restrictions list.
There's no way that a smarter person hasn't figured it out.
Anonymous at Tue, 18 Jun 2024 09:08:17 UTC No. 16240752
>>16240555
First patent, then get a job at a company and do it.
The company would just steal your patent.
Anonymous at Tue, 18 Jun 2024 10:48:12 UTC No. 16240839
>>16240555
Have you tried to train your model yet? Things that might seem "novel" and "groundbreaking" often just end up being 0.1% improvement. Also, what technology, KANs?
Anonymous at Tue, 18 Jun 2024 10:52:39 UTC No. 16240843
>>16240653
Tbf anyone who talks about "AGI" outside of marketingspeak doesn't know shit about AI research, humouring the schizo is hilarious however so I will do that.