Multi-Turn Human-LLM Interaction Through the Lens of a Two-Way Intelligibility Protocol
Harshvardhan Mestha, Karan Bania, Shreyas V Sathyanarayana, Sidong Liu, Ashwin Srinivasan
TL;DR
This work addresses the challenge of achieving meaningful, mutual understanding in human–LLM collaboration for data analysis by implementing the two-way intelligibility protocol (PXP). It models interactions as finite-state machines with tagged messages (RATIFY, REVISE, REFUTE, REJECT) and evaluates a practical human–LLM interface (PEX agents) on radiology and drug-design tasks using a blackboard scheduler. Through controlled simulations and uncontrolled human-subject experiments, it shows that longer interactions increase one- and two-way intelligibility and that machine performance can improve as intelligibility grows, with differences across tasks and experimental settings. The results validate two-way intelligibility as a design principle for human–machine systems and provide open-source code to foster further exploration and application of the approach.
Abstract
Our interest is in the design of software systems involving a human-expert interacting -- using natural language -- with a large language model (LLM) on data analysis tasks. For complex problems, it is possible that LLMs can harness human expertise and creativity to find solutions that were otherwise elusive. On one level, this interaction takes place through multiple turns of prompts from the human and responses from the LLM. Here we investigate a more structured approach based on an abstract protocol described in [3] for interaction between agents. The protocol is motivated by a notion of "two-way intelligibility" and is modelled by a pair of communicating finite-state machines. We provide an implementation of the protocol, and provide empirical evidence of using the implementation to mediate interactions between an LLM and a human-agent in two areas of scientific interest (radiology and drug design). We conduct controlled experiments with a human proxy (a database), and uncontrolled experiments with human subjects. The results provide evidence in support of the protocol's capability of capturing one- and two-way intelligibility in human-LLM interaction; and for the utility of two-way intelligibility in the design of human-machine systems. Our code is available at https://github.com/karannb/interact.