LLMs are not Artificial intelligence

LLMs are not Artificial intelligence LLMs are associative memory, with specific input and output modules. LLMs exhibit many of the needed characteristics of associative memory. The main one being the ability to connect seemingly distant concepts. Associative memory is a prerequisite for reasoning, associative thinking and imagination. However intelegence itself requires more than associative memory and goes beyond reasoning, an appearance of thinking and imagination. biological imperatives There is a lot of talk about AI and AGI, but what is it really? The word ‘language’ in large language model (LLM) includes non-text based languages, visual language and spatial languages as well. As large language model (LLM) expands its range of skills, to include extremities such as hand and legs will show its skills in 3D movement and agility. There is no reason why a LLM can expand into significant 3D motion and skills. As large language models (LLMs) expand their #3D spatial spills such as planning, and navigation Let’s take a hypothetical example. We will build a LLM dog. This will begin with the connection of the appropriate actuators as the output level. This will give the required leg management s well as the addition of visual and vocal sensors and actuators. This artificial creature dog would have the expected mobility and skills to mimic a natural dog’s movements, and communication abilities. With this dog platform the LLM would then learn all the customary dog movements and responses. To move past the ‘prompt’ stage of a LLM the dog will have built in prompts or needs including eating defecation, relationships, play and more. Given extensive training in the form of videos, ‘play’ and social interaction the dog would after a relative short time have many of the behavioral attributes of a natural dog. This this would establish one form of GAI putting a face on the concept of AGI at one level, the elvel of a dog, dog intelligence. The next stage would be to construct another AGI animal, say a pig or a chimpanzee. These cases each would establish a new level of AGI, pig level AGI, and chimp level AGI. The advantage of this development, beyond the actual mechanical and AGI technology it would provide a road map for classifying levels of AGI. The last stage would be the humanoid stage; here the humanoid AGI would mimic all the physical agility of a human without any pre-programed guidelines. All these AGI levels would establish benchmarks for classifying levels of AGI. This scale of GAI would have its own correspondence and would go beyond of the autonomous vehicle analog categories. The expanded goal matrices beyond a destination and the inclusion of inherent conflicts in motivation itself would for a major area of research. The existing AGI categories are human centric and leave specific notations of ‘skills’ as an open question. These SGI creatures would all be at the equivalent of vehicle level 5: full Automation. Or using MRM’s principles offer a material basis when gauging the proficiently at each level. On additional benefit would be to lessen the distortions that the present human centric perspective involves. This line of research would offer a concrete area to develop ethical references with regard to the levels of AGI Still waiting for AI There are addition areas of knowledge advancement. Some of the interesting advances are the development of the self-prompt or motivational GAI development. From the dog to the humanoid, the development of motivation as the central management theme, replacing any pre-planned schemas. At present it is obvious that a expansive associative memory is an important component of intelligence but actual intelligence requires some as yet unknown components. The popular concept of a Savant is i