BLUE & ORANGE blog_e

neuroscience, consciousness, AI

The semantic understanding TAUTOLOGY hypothesis, which should underlie AI research.


Considering semantic understanding:
There is a majority opinion that 'AI does not have semantic understanding.' I think there are people who fear the emergence of AI that is similar to human. My consciousness hypothesis shows also semantic understanding. For example, considering minimum semantic understanding, if understanding a red apple at first, my hypothesis understand 'red is apple' and 'apple is red' only.

It is said that is tautology. But my hypothesis shows that the first semantic understanding is starting from tautology. 'A yellow apple' or 'the things that is red but that isn't apple' should be after learning more. We shouldn't simply say that tautology means ununderstood negatively. We should think whether tautology is negative for semantic understanding truly, or not.

Considering science, tautology is treated as a symbol of ununderstood. For example, when asked why thing burns, someone can answer 'it catches fire'. But it would be said that is tautology, because when asked why 'it catches fire, someone can answer 'thing burns'.

In the meantime what do you think if answered there is oxidation reaction? Possibly many people would think you understand why thing burns, even if you only know 'why oxidation reacts' as 'thing burns'. Actually knowing 'thing burns' and knowing 'oxidation reacts' should be the same level understanding if you know each only. What is the difference? In this case, that is whether it is scientific or not. And whether it is as same as the thinking as observer's thinking and whether observer can sympathize with your associative memory, or not.

[Hypothesis]:
That is, to receive empathy from observer, here you should make an answer based on considering observer's hope. In the meantime, in reality usually you should be enough to answer your associative memory since semantic understanding is starting from tautology.

So here at this point, for answering this question with social view point accordingly semantic understanding is the definition which observer decides. So you need to infer what observer hopes is, what kind of answer, scientific or not, or so, as well as simple associative memory. You can remember an old memory that you are happy to be complimented by others, or you are criticized to answer 'it catches fire'.

With scientific view point semantic understanding should be thought by kind of information theory. And in reality the problem of the semantic understanding which we should pay attention is not tautology. We should be at the first step to understand what the semantic understanding is, and shouldn't delete the possibility of kind of tautology.

In the field of education also it is said that 'learning with hearing' and 'semantic understanding' is different. But from the point of neuroscience there is a possibility of minimum 'semantic understanding' if there is a memory (information element) which can associate. (In the field of education I can understand we should make an answer based on considering observer's hope.)

That's it. This semantic understanding tautology hypothesis is a part of my consciousness hypothesis. Considering minimum semantic understanding, my thought is considering association between information element induces clarification of semantic understanding as same as minimum consciousness. In the minimum semantic understanding information element (memory) should be minimum. And it may be seemed to be tautology. But I would like to share that common understanding that tautology does NOT mean ununderstood. (Of course considering from minimum to further scope there are depth (quality) and width (quantity) viewpoint for the semantic understanding.)

I think @Numenta HTM thought it had semantic understanding naturally based on having intelligence. The thought is similar to my hypothesis in the viewpoint memory underlies semantic understanding. I would like @DeepMindAI and @demishassabis to research also semantic understanding.

My consciousness hypothesis is shown in this link:
The new hypothesis of consciousness mechanism. - Consciousness toy model program was released. - BLUE & ORANGE blog_e

Attached is the illustrated concept of my consciousness hypothesis. And as you can see, semantic understanding is included in it.
f:id:mambo-bab:20141206224806p:plain

This entry is retouched based on the twitter on 5/1/2018.