Agentic AI and environmental responsibility (Part I)

 Today, we present a research paper regarding Agentic AI (Artificial Intelligence) and environmental responsibility.

 

For Christadoss and Panda (2024), in their paper “Harnessing Agentic AI for Sustainable Innovation and Environmental Responsibility”, environmental damage and climate change are accelerating. As a result of this reality, new tools are needed to help the fight against this reality: agentic AI can be that new tool.

But what do the authors consider to be Agentic AI?

The authors invite us to “think of it as AI that doesn't just analyse data but takes intelligent, independent action”. In other words, Agentic AI are systems that perceive situations, make actions, carry out complex series of steps autonomously to obtain a certain objective, that can be sustainable. These systems correspond to a group of AI agents, with several applications, for example, the management of complex energy grids, the discovery of new materials to capture carbon, the creation of more efficient supply chains, and the vigilance on our natural world. In fact, these systems after the reality’s perception can physically or digitally carry out a whole plan from start to finish.

 

But are there any environmental burdens regarding Agentic AI?

The main environmental burden of Agentic AI is the massive computing power needed to the operation of this technology and its training. Furthermore, there can be difficulties due to computational bias and control, being the partnership with humans the key for an output sustainable and resilient.

The main question is whether the AI’s operational emissions are outweighed by the savings it creates. In fact, the training of a single large AI model can emit more than 284,000 kilograms of carbon dioxide and to this value we should add value due to the continuous operation. Consequently, the data centers that house Agentic AI are becoming a significant part of global electricity demand.

We as users should take this reality into consideration when using these tools, because not only is electricity consumption important but also due to the efficacy we would like to have when using them.

 

[This text continues]

 

Enjoy.