You could have a "cognitive architecture" with LLM-type components or soft modules. Neural nets and "cognitive architectures" are not mutually exclusive.
These guys should team up with Verses AI. Similar approach but appear to be a few years ahead and have to protocols in process of going through the IEEE. Also rolling their beta agent out this year.
This research will yield some interesting results. On the other hand, we already have many units of RGI- real general intelligence. Some of them are quite good.
Peter, it’s rare that I come across someone that make sense to me. The last one was Dr. Michael Levin. I assume you know of his work. I’m also a big fan of Elon and own Tesla stock. However his approach to FSD is one of brute force. I hope he is successful but it is taking a long time and I fear he might fail. I’ve been thinking if someone does create true AGI it would make Elon’s approach almost immediately worthless. If he knows of what you are doing I certainly wish he would send some money your way. As you say AGI is a trillion dollar business. Actually I’m sure it is many trillions.
What can this AI with a brain do, that generative AI cannot? Can it write code? Can it solve problems? How does it score on the suite of IQ tests that are commonly used to compare generative AI system IQ?
@@benoitmayaux5063 we've got a world full of usable tech in the form of software - hundreds of millions of users are finding LLM's so usable that there's a stampede of investment to make it 10x, 100x, 1000x more capable and useful. I'm interested in if this explainable technology can at least compare with, say GPT-3 and that it's scalable.
Im really getting tired of all these discussions centering around defining or redefining terms. Oh yeah, real smart, ur theory can always be right if you just define words so that everthing fits with your ideas