Google's emissions are increasing more and more.It's the fault of artificial intelligence

Over the past five years, Google's emissions have increased by almost 50 percent due to artificial intelligence.

Generative artificial intelligence continues to interest, to obsess, we might say, technology companies, such as Google, even if many questions remain open that could impact their future.Between all, the most pressing is the economic one, with more and more voices, including the influential Sequoia investment fund, which they underline the lack of economic income (in the face of billions of investments).In short, these systems cost a fortune and there is not yet a business model capable of supporting them economically.

Among the heaviest voices for companies, in addition to that of the sector's qualified and overpaid employees, is the energy bill, as Lifegate has already reported.In fact, each interaction with a chatbot or generative AI consumes a significant amount of energy, which increases in cases where users ask to generate a sound, a song, an image or a video.

Google's net-zero emissions goal is in jeopardy

Let's multiply everything by the millions of people who now have access to similar tools and the problem stops being corporate and begins to have a systemic dimension, as well as environmental.In the new environmental report of Google, for example, the company declared that it increased its emissions during 2023 by thirteen percent compared to the previous year, reaching a total of 14.3 million tons of CO2;a quantity that increased by 48 percent compared to 2019.

The report cites other numbers and makes great promises but this single piece of data is enough to give an idea of ​​the impact the generative AI of the technology that Google has adopted has had on the sector, even with the services discussed like AI Overviews.Yet, for many years, the so-called Big Tech companies have been the most generous in investments in clean energy and plans to become carbon free.Not anymore, so much so that Google is forced to admit that its "ambitious plan" to become a zero-emissions company by 2030 is in danger:“It won't be easy,” the document admits bitterly.

A disaster, in short, which "was mainly due to the increase in energy consumption of data centers and supply chain emissions", continues the report.The company does not hide what caused this peak in energy consumption (and therefore emissions), writing that "the reduction of emissions could become difficult due to the increasing demand for energy resulting from the increased processing intensity of artificial intelligence.”

The problem obviously doesn't just concern Google. Microsoft (allied to OpenAI), Half And Apple they all have big plans in this sector, and they all have had big environmental ambitions, with programs similar to those of Google.The concrete risk is that the promise of becoming carbon neutral by 2030, rather common and widespread, is forgotten and disregarded, sacrificed on the altar of the race for AI at all costs.

However, the environmental issue of AI is not limited only to the energy that it consumes:data centers indeed need enormous quantities of water, which is mostly used for systems cooling.According to data cited by NPR, the average data center would consume more than one million liters of water per day, or as much as a thousand US homes.

Unless there are revolutionary technological innovations, the entire sector is at a a crossroads:

• to cool the equipment they can consume more water and "save" electricity, which is still used to power the data centers,
• use less water and consume even more energy.

None of the choices are optimal and on the horizon there is no third way in sight or a miraculous solution.In short, perhaps it is time to add the crazy consumption of environmental resources to the list of ethical problems, such as systematics violation of copyright, linked to AI.In short, before generating a funny image with MidJourney, it's better to think about it.

This article was corrected on July 7.Google's CO2 emissions in 2023, in fact, were equal to 14.3 million tons and not 14.3 tons of CO2 as initially written.