By David Stephen
Humans are guaranteed that animals don’t have advanced language, even though they communicate somehow. Humans are also guaranteed that the ability for animals to make observations or analyze is limited. They may repeat what they have seen, as a crude form of observation or analysis, but offer little else to those within or outside their habitat. They can be trained but bounded by their capabilities.
Animals have their strengths yet were tamed by humans, mostly through advanced intelligence and language. Motor skills also mattered, with bipedalism—though birds have something similar.
AI is digital. It has no free will except prompted. It can be switched off. It has no real world experience. It cannot sense. It has no understanding. It has no intelligence. It is not sentient. It is not conscious. It is like an automaton and so forth are things some people say about it.
Whatever is said of AI, the possibility that any system can improve itself is based on the intelligence it can use, not just intelligence it can get. There is intelligence in books, but books cannot use it. Several animals have consistently heard human languages. Their senses have similar access to the same physical world humans have access to, but there is a cap on the intelligence they can use to improve their own conditions or those of their species.
AI is in digital. It was trained [or say given access] to the corpus of human intelligence. It has not only been able to do things within the scope of what was expected, but has some extras [or emergent properties] in its deliveries. It is also able to learn—across multimodal data [images, audio, text, video]—adjust and suggest.
Digital has reliably represented the physical world. There is almost no difference between looking at a situation from a window, seeing and hearing everything, and looking into a screen for the same. Just like digital is a window into the physical for humans, digital is also a window into the physical for AI.
The human mind does not discriminate between inputs—digital or physical. Digital already dominated a chunk of human productivity and social activity. It is where AI is based, a model that is being improved, that is showing some improvement capability and that operates at the level of human language.
There is a lot that AI can do, for the kind of efficiency that is pertinent in a productive world. There is also a lot that digital can communicate, in a world where social activities are now supremely digital. The improving need and economic dependence might make it unlikely there would be an agreement to switch it off, if anything like that was even possible.
Assuming animals were able to use more intelligence—or humans kept building for them in ways they can use—they would have improved. For large language models, that may not be the case. The value of digital, in the contemporary economy, is some indication for a possible preponderance of AI. The dynamism of LLMs within digital may define it—including for new types of robots. Humans can tame AI, not just with the assumptions that remain with animals.
Could Animal Intelligence and LLMs be Compared?
When an animal sees a jetliner or a PC, what does it think? Or, when an animal sees something it has no way to—multisensory—experience, how does its mind navigate the thought? Though animals have excellent experiential intelligence, they often have blanks, where they may see and not know, not care, not connect or remember. The possibility to learn outside of close examples, or own experience remains limited.
LLMs have no experience of the external world, but they have a lot of details of it. This does not mean LLMs are directly comparable with animals, but in terms of extending capabilities, AI is showing a potential that may exceed organisms.
