Leveraging Hybrid Intelligence Towards Sustainable and Energy-Efficient Machine Learning
Daniel Geissler, Paul Lukowicz
TL;DR
The paper tackles the environmental impact of large-scale AI by arguing that energy efficiency is often undervalued in model development. It proposes a Hybrid Intelligence framework that combines HITL visual analytics with LLM agents to surface inefficiencies and guide energy-aware training decisions. Key contributions include integrating comprehensive energy-tracking with explainable visualizations and outlining an evaluation plan centered on HAR datasets to quantify potential energy savings. The work envisions a lifecycle-oriented toolkit enabling energy-aware optimization from idea to deployment, with practical implications for reducing resource consumption in real-world ML pipelines.
Abstract
Hybrid intelligence aims to enhance decision-making, problem-solving, and overall system performance by combining the strengths of both, human cognitive abilities and artificial intelligence. With the rise of Large Language Models (LLM), progressively participating as smart agents to accelerate machine learning development, Hybrid Intelligence is becoming an increasingly important topic for effective interaction between humans and machines. This paper presents an approach to leverage Hybrid Intelligence towards sustainable and energy-aware machine learning. When developing machine learning models, final model performance commonly rules the optimization process while the efficiency of the process itself is often neglected. Moreover, in recent times, energy efficiency has become equally crucial due to the significant environmental impact of complex and large-scale computational processes. The contribution of this work covers the interactive inclusion of secondary knowledge sources through Human-in-the-loop (HITL) and LLM agents to stress out and further resolve inefficiencies in the machine learning development process.