Winning at LLMs
Thoughts on how to win AEO / AIO (the new SEO). LLMs, for all the amazing things they can do are still at their core a prediction model for “what word comes next”. Sure, it’s tokens, not full words, but the point stands. As more models approach full ingestion of all available text data from humanity, all the flavour and nuance between models comes from weighting and fine-tuning. Exactly how training data is weighted is proprietary in most models....