Chat GPT is a powerful AI used by more than 180 million people. Having such a powerful and big machine costs a lot to build, and people cannot determine how much it costs to create this machine or how much it costs to maintain it. Many things are needed to help keep chat GPT running and functional. Engineers use a lot of water to cool down chat GPT. In the last environmental report done, consumption spiked up 34% in water usage from 2021 to 2022. They use precisely 1.7 billion gallons of water each year, and this is growing. Almost all of this growth in water usage has come from AI development. Water usage isn’t the only impact that chat GPT has, it also has a very high usage of carbon dioxide According to estimates, ChatGPT emits 8.4 tons of carbon dioxide per year, comparable to the usage of 5 cars for a lifetime. The energy used to train chat GPT is also massive. It is equivalent to a normal house energy consumption for 700 years.

Chat GPT has high energy usage and energy consumption from the training phase, inference phase, data centers, and electrical waste—these phases of the program chat GPT. The training phase runs data through the AI to make it more accurate. In the inference phase, the AI receives computational resources to process user input and teaches how to generate responses. After these two processes, the AI is ready to use functionally. Finally, the data centers of chat GPT use a lot of energy, which comes from the storage of the data required to run chat GPT. It uses fossil fuels and renewable energy. Even though many companies use renewable energy, a handful use fossil fuels. Chat GPT data centers get very hot and need to be cooled down, so they not only use lots of water but use even more energy to cool down by having a cooling system that is not only not energy efficient but is often wasting energy by running even when the cooling system is not required

 The amount of resources used for one AI is huge, but there is not just one AI and many of these AIs consume the same energy. This energy and carbon consumption demands that a more sustainable AI be built that doesn’t require as much energy consumption, for example, having an AI that can run off just natural energy. While chat GPT is using all this energy, it is still not working as well as it could. For example, chat GPT does not properly comprehend some things and ideas as we humans do; it uses data to understand emotions. It has many limitations on emotions and sensitive topics. Another problem with chat GPT is that it often needs more accurate information about specific situations and depends on lots of input to get an exact answer. Many issues are common for chat GPT even though it uses many resources to keep functioning.