By Sanjay Sehgal, Chairman & CEO, MSys Technologies , Venture & Angel Investor, Philanthropist
This is not yet another article dictating the impending doom of climate change and its effects because we are past that precautionary tale. The climate crisis is real and the pace at which we adopt change to curb it is not fast enough or we are not scrutinising the impact of some of our most used tools & technologies. One such instance is of AI, a single GPT query consumes 15 times more energy than a Google search query. It consumes 500 ml of water to generate roughly 40 prompts. AI has a significantly more high water and energy consumption over your search engines and rightfully so to match the innovation. So asking random questions to ChatGPT has serious consequences to the environment, if not through the plagiarism done using it.
While it is hard to ascertain the exact impact caused by the AI systems, it is no surprise that the development, maintenance, upgrade, and eventual disposal of AI technology all come with a large carbon footprint. The urgency to address this is crucial because as datasets and models evolve, the energy required to train and run AI models also rises which has then a significant influence on greenhouse gas emissions and eventually aggravates climate change.
Billion-dollar problem or solution in making?
There is strong advocacy from tech leaders globally on decarbonising the industry with the help of tech and AI-based solutions. For instance, Google DeepMind. Google’s DeepMind is leveraging AI to optimize energy consumption in data centers and reduce cooling costs by up to 40%. The Nation Grid ESO, electricity system operator for Great Britain, also uses AI to improve the accuracy of its forecasts of demand for electricity, enabling better integration of renewable energy. The use case of AI to save the planet is also evolving.
UNEP’s world environment
Situation Room (WESR) which was launched in 2022 is developing a centralised system to analyze huge chunks of data to predict CO2 atmospheric concentration, changes in glacier mass and sea level rise. Using AI to solve climate change is also ironic considering the carbon footprint of AI itself. A 2019 study called The MIT Technology Review indicates that a single AI model can emit up to 6,26,000 pounds of carbon dioxide equivalent to five average American Cars’ total lifetime carbon dioxide emissions.
Who will bell the cat?
A collaborative and transparent approach between government and AI companies to create independent industry bodies that have representation & objectivity is the need of the hour. Proactive measures suggested by the body should help create a framework for judicial laws while promoting innovation that helps create clean energy solutions.
On an immediate basis, there needs to be more conversations and awareness generated around the urgency of the situation. Transparency is a tricky road for tech companies but with proper channels of communication and intensive pressure to be eco-conscious, there is hope to yield AI as a tool and not a weapon against the environment.