Studies show real behavioral profiles beat synthetic personas in fine-tuning, while new work examines empathy control, semantic consistency and mechanistic interpretability.
ServiceNow's research shows that behavioral profiles extracted from real interactions offer tighter personalization than synthetic personas. The method combines supervised fine-tuning and multi-perspective reasoning, shifting focus from identity temp...
A ServiceNow framework extracts high-fidelity user profiles from anonymized social media posts. The behaviorally grounded profiles improve base models and outperform synthetic personas in both fine-tuning and multi-perspective reasoning, shifting how...
A study on collaborative conversations shows that the semantic similarity of generated replies changes with model and chat history. Prompts and context alone cannot preserve consistency; infrastructure and design strategies are needed for stable, com...
A study on three instruction-tuned LLMs shows that a decodable empathy direction produces only partial shifts in automated scores. The affective control passes, but the cognitive instrument is too coarse. Gemma's Recognition ablation lowers the class...
Silico brings mechanistic interpretability tools previously reserved for elite labs to researchers and startups. The bet is that understanding models from the inside makes them safer, more controllable, and better suited to local deployments.
