Their study reveals that a smaller language model can achieve brain alignment comparable to larger models, offering a more efficient, accessible approach to advanced AI