Description of the project

COOMEP is an interdisciplinary research project led by FARI, that aims to develop knowledge and expertise that enable households to coordinate their energy usage behaviours, based on AI proxies, that is, autonomous decision-makers selected and configured by their users to act on their behalf.

The project is structured around four key areas:

Monitoring and Modeling Household Energy Consumption

This part of the project focuses on tracking household energy usage through high-tech devices. By analyzing real-time data with algorithms, it identifies which appliances are used, when, and for what purpose. The goal is to to create a model that is qualitative in nature about the household consumption of energy, offering detailed insights into household electricity use.

Smart Coordination of Household Energy

This part of the project explores how household energy production and consumption contribute to collective resource use. A multi-agent model is developed to determine when coordination mechanisms should be introduced. The goal is to optimize energy consumption through ADM proxies, which will soon be commercially available to manage household energy needs. These proxies must identify the best times to buy or sell energy. Recommendations will be made on managing this ecosystem while ensuring, in collaboration with the legal team, that human rights are protected.

User-Centered Energy Management

This part of the project integrates the perspectives of people who will interact with AI proxies, ensuring their needs are considered. End-user participation is key, as they are best positioned to define how the system should account for their daily lives. For example, for families with a newborn, the very fact of having such a life change will have repercussions on their use of energy and their everyday strategy of energy consumption. Social science methods will be used to have AI experts rethink their ‘models’ or consider alternatives that take into account these other needs.

Safeguarding Fundamental Rights in AI Systems

This part of the project develops a precautionary approach to prevent potential infringements of fundamental rights. It focuses on understanding how design decisions involving AI proxies could impact our rights. For instance, the fine-grained description of specific households’ energy use may have far-reaching effects when that description is used as the basis for the creation of mechanisms that would apply to other communities. Even if simulations are used to mitigate potential misrepresentation, the assumptions behind these models must be explicit, as they influence the behavior of smart agents. This section aims to address these concerns and ensure that the design decisions do not violate fundamental rights in different deployment scenarios.

Objectives

The overall goal of this project is to co-create a socio-technical distributed coordination mechanism based on proxies that is fair towards its users and the community, while not being manipulative, enhancing rather than diminishing the agency of those concerned.