The training of AI / Large Language Models on aggregated human-generated data and their design for statistical consistency result in outputs that typically align with the most common (neurotypical) human responses. They lack the unique cognitive patterns and atypical neural connections that characterize neurodivergent thinking in humans.