LLMs are the future. How do you think? Through graphs, or through texts? So why build an AI model that isnt a substitute for how we think?

I do see the great potential, like wolfram alpha is an amazing software paired with GPT it can produce amazing results. And I think in the future AI will utilize the models as tool, just like it already is with ChatGPT plug-ins. We gave AI a voice, we let AI see and now AI can use tools.