I had this thought last night as we as humans have knowledge of various fields such as psychology, current affairs, technological advancements and growing need of new advancement for better living. We can combine ideas across multiple domains and that happens by just random thoughts. I think to achieve that in a machine we just need 2 solid machinism. One which can produce random images or videos totally random across multi domain and second one who just figures out what it is.
It can be sound simple and it is but the two mechanisms can be any number of LLMs like there can be number of LLMs for image and videos generation, one more thing they must be able to produce this without any prompt or telling them explicitly, and there can be number of LLMs or small LLMs which can part of a greater LLM who just see the end result or summary produced by the smaller LLMs who are smart in one domain such as Biology, Psychology, Physics,Computers, Economics etc. where each will figure out about the video or image based on their own trained parameters and at the end a grater LLM just connect what just have been summarized by those small LLM. Now if we want we can also acquire more than one Greater LLM and connect Ideas by our own and have a soup of our own.
Well this was just my thought and I want to share it with people so I did. Happy to know your thoughts on it. Thank you for reading
andsoitis•33m ago
Not all concepts can be represented visually.