D4RT generalization and Thinking

Thinking in brain is analogous to learning of the AI model, that is, when thinking, due to synaptic plasticity, brain is rewiring and making new changes, which also changes the state of its weights. So when training AI model, perhaps its just reacting when it is not trained, and does not think creatively like human because it does not traverse its weights in its weight space to allow generation of new ideas or thought, and so just reacts to the stimuli.

In AI field, models are trained using data, using encoder and decoder, encoder forms the representation, and to understand it , encoded represents a form of perception or feeling that exists within, i.e. analogue of it in information, which can be abstract solid concepts, from which definitive things can be extracted.

From example of the D4RT 4D reconstruction, information is projected into a space, and from which relevant information is extracted. https://arxiv.org/pdf/2512.08924

For example, if the network is trained on human anatomy model, by given certain range of parameters such as exercise and food intake , it is possible to estimate other parameters such as glucose levels subject to condition. But this forms the higher view with much dependency on many other internal factors. But basically makes it a simulator of a system. i.e., we are relying on complex thoughts that we get, same thought complexity information emerges in such deep networks / generative etc, (as whatever it is called doesn’t matter, but how the information flow is governed by the network, which is ultimately implemented on a physical structure (run) on chip (GPU).