Artificial intelligence-based "80s trend" brought environmental costs back to the agenda

On social media, users transform their own photos into 1980s hair, clothes and photography aesthetics with artificial intelligence tools, asking "How would I look like in the 80s?" The current has brought the need for electricity, water and other resources behind the images created in a few seconds to the agenda again.

According to the AA correspondent's compilation from the "Environment and Health: Carbon, Water and Land Footprints" study of the United Nations University Water, Environment and Health Institute, the energy consumed for a typical image created by artificial intelligence is enough to run a 10-watt LED bulb for approximately 17 minutes.

The water footprint of the electricity consumed for the production of an artificial intelligence image is calculated as approximately 29 milliliters. Although this amount corresponds to approximately two tablespoons of water for a single image, it reaches 29 million liters in the production of approximately 1 billion images.

According to Google Trends data, searches for the phrase "80s trend" increased by 5 thousand percent last week compared to the previous week, while the relevant artificial intelligence image was shared by millions of users on Instagram.

The environmental cost of artificial intelligence does not only consist of electricity consumed during image production. The production of electricity required for the operation of data centers, cooling of servers, establishment of infrastructure and production of the hardware used are also among the elements of the total environmental footprint.

As the scale of production grows, the need for resources also increases

According to the calculation made in the study, global data centers consumed approximately 448 terawatt hours of electricity in 2025. The International Energy Agency (IEA) expects the total electricity consumption of data centers to increase to approximately 950 terawatt hours in 2030. The IEA states that one of the most important drivers of this increase is the proliferation of artificial intelligence.

According to the study, it is predicted that the water footprint associated with electricity consumption in 2030 of data centers supporting artificial intelligence may reach 9.3 trillion liters and the land footprint may exceed 14 thousand 500 square kilometers. These figures do not directly belong to the "80s" trend, but express predictions about the growth of artificial intelligence infrastructure on a global scale.

Most of the energy consumption of artificial intelligence occurs after training. According to the research, processing daily requests that users send to the model accounts for approximately 80 to 90 percent of total AI energy use.

The research calculated that a typical artificial intelligence image requires approximately 1450 times more energy than the basic text classification process, while according to the same study, a typical image production consumes approximately 2.9 watt-hours of energy.

Energy consumption varies depending on the model in image production

According to a study by Hugging Face, an open source platform where artificial intelligence models and data sets are shared, and Carnegie Mellon University researchers, an average of approximately 2.9 kilowatt-hours of electricity is consumed per thousand requests in image production.

The study also reveals that the energy needs of artificial intelligence tasks vary significantly depending on the model and task used.

Therefore, instead of giving a fixed electricity consumption for a single artificial intelligence image, the model and calculation conditions used need to be taken into account.

The need for resources continues on the hardware side as well. It is predicted that artificial intelligence-generated electronic waste may reach 2.5 million tons annually in 2030.

Thus, although a retro photo created in a few seconds on social media does not constitute a huge resource consumption on its own, it is added to a broader resource chain that includes electricity, water, land and hardware needs as part of billions of artificial intelligence processes.