How Good Is Fable, Really? And the Intentional "Poisoning" of Your Deep Research
Introduction
Full credit goes to the original author, linked below. All blog posts were reposted either with permission of the author, or by anonymous submission by SAIRC members like yourself.
Most reviews of frontier models focus on coding, so the author tests something different: can Fable 5 read customer interviews and surface the patterns, themes, and nuance an expert human would find? The experiments lead into a second, cautionary thread about how deep-research outputs can be deliberately skewed.