As a non-expert, I believe that the performance of generative AI will continue to improve and that the purchase of chips will undoubtedly increase in the future.
However, one thing to consider is that the functional growth of generative AI cannot proceed exponentially from its current level, and it may have already reached the upper limits of its growth curve.
From an investment and business perspective, it is still in the early stages and will certainly expand further. Nonetheless, in terms of AI performance improvement itself, I believe we are now on a well-visible path.
How about an AI model that is somewhat slower but prioritizes abstraction and structuring, effectively building a kind of multi-layered filter? I'm talking about a biased AI model.