By: Robert A. Kobek, RRP, President

Mobius Vendor Partners

A very smart and well-intentioned client of ours, that’s who. “Has there been any consideration on the environmental impact of utilizing AI programs?” Brilliant because we were stumped. And that doesn’t happen very often. We try to take a wholistic 360-degree view of everything we do. One of our internal codes is to control the variables. This one question is one of the most intelligent questions we have been asked. We asked what we thought was every question that could be asked about AI, including but not limited to: security and accuracy.

CustomerCount is very glad to offer an AI-based integrated text analytics/sentiment analysis solution to our platform. Keatext© (Keatext | AI-Based Text Analytics) has been our partner since before AI was cool.  Providing enhancements for our clients that range from intelligent machine learning to interpret and report on unstructured data to structured reports that included focused recommendations for reputation management. The reports are integrated into our reporting portal through an ‘iframe.’

So, we have skin in the game

Does AI have an environmental impact? We know bit coin sure does. It sucks in massive amounts of natural resources and elements to produce enough energy to power the servers necessary to manage that currency. It’s like a whale swimming up a river – it pulls a lot behind in its wake.

The electricity demands of data centers are one major factor contributing to the environmental impacts of generative AI, since data centers are used to train and run the deep learning models behind popular tools like ChatGPT and DALL-E. This puts a huge strain on municipal utilities including water supply. For example, Amazon has more than 100 data centers worldwide, each with 50,000 servers. You read that right, that is more than 150,000 servers, each having very dense requirements to operate.

And here’s a kicker, for those utilities to manage that much absorption, they use diesel power generators.

Here is our answer from Charles Cosimard, President and founder of Keatext, “our systems do not require the use of server farms or any server structure than any other business of 15 people. We are not reliant on any outside source of hosting other than what any business might require, and we provide all our services in-house. So, as for any additional environmental impact other than normal technology we are very glad to say we are very sustainable”. Adding;” we run our proprietary models on single GPUs that are hosted on AWS. They are configured to run only when data is required to be processed (the rest of the time, machines are off). We consume some additional GPUs when we train our models, but at our scale, it’s only a few hours of computing time every month. We’re a 100% remote work model, with 15 employees, and our only physical assets are laptops that we provide to our employees.

So, we are good citizens with secure processes, compliant wherever required and environmentally sound!

Explained: Generative AI’s environmental impact

Rapid development and deployment of powerful generative AI models comes with environmental consequences, including increased electricity demand and water consumption.

Adam Zewe | MIT News
Publication Date:
January 17, 2025

Explained: Generative AI’s environmental impact | MIT News | Massachusetts Institute of Technology