Artificial Intelligence or Negligence?

Alyna Sigel

It’s easy. Don’t want to spend several hours writing a paper for class? Copy and paste the prompt into Chat GPT and it's done. Can’t find the answer to your question online? Ask AI and your problem is solved. Don’t understand how these complex artificial intelligence systems work? No worries, it’s easier not to think about it. It’s easier, but is it better? 

There are many concerns surrounding the increased use of artificial intelligence ranging from ethical concerns to fears of job loss and autonomous weapons. While there is an abundance of discussion about whether or not AI is a threat to society and humanity, many institutions, such as universities and businesses, have decided to adapt to this new, useful tool. It is understood as a technological advancement that is unavoidable, so instead requires adaptation. However, changes in syllabi and job assignments to accommodate the use of artificial intelligence do not assist in countering one major threat posed by the technology: the threat to the environment.  

With an extreme carbon footprint, the disposal of electronic waste, and the overuse of water, artificial intelligence may have more negative impacts than positive. By first focusing on the carbon footprint, it is necessary to understand the energy use needed to train AI. The amount of power needed to train AI models has doubled every 3.4 months since 2012. By training an AI model, about 626,000 pounds of carbon dioxide are produced. For reference, this is equivalent to flying round-trip between New York City and San Francisco about 300 times. It is predicted that by the year 2040, the Information and Communications Technology industry will be responsible for 14% of global emissions. 

The creation and use of AI produce e-waste and a lot of it. The World Economic Forum projects that by 2050, over 120 million metric tonnes of e-waste will have been generated. This e-waste, containing hazardous chemicals such as mercury, lead, and cadmium, imposes significant threats to the environment and human health. As computing devices are frequently replaced, advancements in hardware technology cause AI to contribute to the already existing e-waste problem.   

The amount of water consumed by artificial intelligence is an additional concern. Data centers, which are essential to AI operations, are increasing in number to accommodate the demand for artificial intelligence. Unfortunately, this means water usage is also growing. While smaller data centers use about 18,000 gallons of water per day, larger centers, like those supporting Google services, average around 550,000 gallons of water per day which is equivalent to the daily water use of about 4200 Americans. AI models consume water during both onsite server cooling and offsite electricity generation. To prevent servers from overheating, data centers often use cooling towers or outside air which requires a great amount of freshwater. Water is also consumed offsite in the process of generating electricity for operating the data centers. 

Simultaneously, AI brings significant hope for innovation in a variety of fields. In the medical field, artificial intelligence can be helpful in diagnosis and treatment determinations. It also allows for advancements in computer science by improving cybersecurity and in engineering by assisting in evaluations of efficiency and safety. Furthermore, the use of AI can lead to improved quality of life for those utilizing it as a tool in their personal lives or evaluating global circumstances and solutions. It is undeniable that there are many benefits of artificial intelligence use, especially as the fields continue to develop. 

However, the reality is that while artificial intelligence is helpful and could have important contributions to solving global issues, it is also harming the environment. Attempting to halt the use of artificial intelligence is therefore both unrealistic and potentially a step back in many fields. However, the goal of mitigating the impacts may be more reasonable. Transparency, responsible AI use, technological innovation, and climate change awareness are all influential in balancing the use of AI with the protection of the environment and can help to achieve this goal of mitigation. 

Transparency is often considered an important first step. Transparency of quantified information allows for environmental impact assessments which can then be used to determine solutions. Currently, AI model developers do not share enough information about the water footprints of the technology. However, this information is not difficult for developers to calculate and would be incredibly helpful if shared publicly.  

So while AI is certainly an easier solution to many issues, is it better? The answer seems to be that it depends. It depends on transparency and it depends on whether or not it can be made to operate more sustainably as a result. AI has the potential to save or ruin the environment, with a thin line between intelligence and negligence. Calling for transparency and responsible use of artificial intelligence could be the crucial difference.

Citations

1. Crownhart, C. (2024, October 28). AI will add to the e-waste problem. Here’s what we can do about it. MIT Technology Review. https://www.technologyreview.com/2024/10/28/1106316/ai-e-waste/

2. Kanungo, A. (2023, July 18). The Green Dilemma: Can AI Fulfil Its Potential without Harming the Environment? Earth.org; Earth.org. https://earth.org/the-green-dilemma-can-ai-fulfil-its-potential-without-harming-the-environment/

3. Patterson, J. (2024). Ways AI can improve our world. Rowan.edu. https://irt.rowan.edu/about/news/2024/10/ai-benefits.html

4. Privette, A. (2024). AI’s Challenging Waters. Illinois.edu. https://cee.illinois.edu/news/AIs-Challenging-Waters

5. Ren, S. (2023, November 30). How much water does AI consume? The public deserves to know - OECD.AI. Oecd.ai. https://oecd.ai/en/wonk/how-much-water-does-ai-consume

6. Thomas, M. (2024, July 25). 14 Risks and Dangers of Artificial Intelligence (AI). Built In; Mike Thomas. https://builtin.com/artificial-intelligence/risks-of-artificial-intelligence

7. Yao, Y. (2024, October 10). Can We Mitigate AI’s Environmental Impacts? Yale School of the Environment; Yale University. https://environment.yale.edu/news/article/can-we-mitigate-ais-environmental-impacts