Introduction: What is ChatGPT?
In recent years, one of the fastest growing technologies has been artificial intelligences.
artificial intelligences, among which one of the most outstanding is ChatGPT,
developed by the company Open AI.
By now, working with AI is the order of the day, at university it is a useful resource for some and indispensable for others, but have we ever really thought about how much we pollute with each query? ChatGPT like any technology involved a large energy expenditure for its development,
and maintenance, which has a huge impact at the level of resource expenditure.
Environmental Impact
How does ChatGPT work on a technical level?
ChatGPT is not what you might call a sustainable technology, but before we focus on what it means for the environment, it is necessary to understand what this revolutionary artificial intelligence is. Chat Generative Pre-Trained Transformer, which is the acronym for this generative artificial intelligence, is a program developed by the OpenAI organization. ChatGPT is just one of the many developments the organization has been working on.
Originally, it was just a chatbot that was developed to improve the customer service provided in the channels of companies and public administrations. The idea was that it would be much faster and more accurate than traditional chatbots, but it turned out that it has much more potential than just managing a business chat.
Today, ChatGPT has become an effective language model that can interact with any human in a natural way and maintain a logical and fluent conversation with any person.
conversation with any person. To generate the text with which it responds to its interlocutor and, thanks to its machine learning, this technology uses natural language processing along with a large database of texts, websites and articles to mimic human style and answer questions.
Energy and resources needed to train and run ChatGPT
For this technology to work, it needs to be fed with a large amount of data. Without it, this artificial intelligence would not even exist.
And to process all this huge amount of information, data centers come into play, as they are the ones most responsible for the emission of greenhouse gases and consume the most energy.
Handling and analyzing this data requires a lot of energy. In other words, apart from data, ChatGPT needs to consume enormous amounts of electricity in order to work. Some studies estimate that just in processing the data, companies using artificial intelligence pollute as much as the entire aviation industry combined. The problem is that more and more companies and users will be using ChatGPT and other AI solutions and, therefore, their use will increase energy consumption.
There are already several estimates on the consumption of current ChatGPT usage and these estimates indicate that this generative AI application emits 8.4 tons of carbon dioxide per year, which is more than twice the amount of carbon dioxide emitted by an individual, which is 4 tons per year.
The study is the responsibility of Cornell University. To develop ChatGPT-3, they note, 700,000 liters of clean fresh water were consumed, which is the same as it takes to produce 320 Tesla Model 3s.
But ChatGPT also consumes water every time it has a conversation with a user, specifically a half-liter bottle of water. If we keep in mind that consumption by the 100 million users who asked ChatGPT a question last January alone, we are talking about really big numbers.
Possible solutions
In order to continue progressing as a society we must ensure that our resources are optimized in the best possible way, and as we saw in the previous points one of its main disadvantages is the expenditure of resources when making queries compared to that spent by Google, so although ChatGPT gives us the information already collected many times you have the same information to one more click using your search engine.
Another major dilemma about the sustainability of ChatGPT is where their energy comes from, in an estimated and typical breakdown of energy consumption in modern data centers that use a combination of energy sources. We can see that a highly polluting percentage is the use of coal and gas, and knowing the great energy requirements, we bet on nuclear energy, which is the most powerful and does not pollute.
Conclusion
Although artificial intelligence and in particular ChatGPT have made significant advances in technology and task optimization, their use has a significant environmental impact due to their high consumption of energy and resources. From the large volumes of electricity required to process data to the water consumption associated with its operation, this technology poses a considerable challenge to sustainability.
The growing use of AI implies an increase in carbon emissions and energy demand, which today relies heavily on non-renewable sources such as coal and natural gas. To mitigate this impact, it is essential to optimize the use of ChatGPT and explore cleaner energy sources, such as nuclear power, which, although controversial, can significantly reduce CO2 emissions compared to fossil fuels. We need to adopt more sustainable practices when developing and running AI if we are to reap its benefits without being consumed by it.