Beyond CO2 Emissions: The Overlooked Impact of Water Consumption of Information Retrieval Models

TL;DR

This paper highlights the often-overlooked environmental impact of water consumption in neural information retrieval models, emphasizing the need to consider water use alongside energy and CO2 emissions.

Contribution

It introduces the significance of water consumption as an environmental factor in neural retrieval models, expanding the scope of sustainability considerations beyond CO2 emissions.

Findings

Water use in neural models is substantial and often ignored.

Water consumption impacts are comparable to energy-related emissions.

Calls for comprehensive environmental assessments including water use.

Abstract

As in other fields of artificial intelligence, the information retrieval community has grown interested in investigating the power consumption associated with neural models, particularly models of search. This interest has become particularly relevant as the energy consumption of information retrieval models has risen with new neural models based on large language models, leading to an associated increase of CO2 emissions, albeit relatively low compared to fields such as natural language processing.