Debate over synthetic data and information limits in large language models
Commenters debate whether synthetic data generated by large language models introduces genuinely new information or merely remixes existing content, and how that affects scaling and reasoning capabilities.
