Search papers, labs, and topics across Lattice.
2
0
5
LLMs can only decompose complex data analysis queries into executable tool-based plans with ~85% accuracy, highlighting persistent gaps in tool understanding and planning ability for contact center AI.
LLM-based contact center QA systems exhibit systematic unfairness, with contextual priming inducing CFR increases up to 16.4%, even in larger, strongly aligned models.
