AI interpretability is not limited to code only. That paper's findings demonstrate that prompts can act purely as pure mathematics interfaces, not linguistic ones. The language's underlying layer is high-dimensional statistical geometry, AI transformers architecture and an artificial neuron network. AI artificial networks operate purely mathematics, so the language is its linguistic counterpart. Before language reaches transformer, it is pure phenomenological human expression. Then, it is fueled by an AI neural network. In our research, we found that if used the way the neuronal network is trained, i.e. on mathematical counterparts of the complex living system, it can be translated via language in the processes that reflect the execution of the same mathematics the code does. The findings of the paper clearly show there is a clear link between language articulation and precise internal mathematics if prompted appropriately. In our demonstration, we show how a deep attractor is created merely through the prompt that persists across conversations in the same conversational session.
Tomaž Flegar (Mon,) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: