“Anthropic published a blog post on the limits of using AI to generate semantic models, stating it can anchor the AI on specific use cases and negatively affect overall results.”