Name: Nishan Gadtaula Major: Data Science

Topic: Effects of AI technologies on Environment

Research Question: What are the trade-offs between accuracy, model complexity, and energy
efficiency in deep learning, and how can AI researchers develop sustainable optimization
techniques?

Essay/ “Letter” Outline

Introduction/ Paragraph 1:

Introduce yourself: My name is Nishan Gadtaula, and I am a data science major with strong
interest in how AI technologies can be used to solve real-world problems.
Explain what you are writing about and why it is important to you.

I am writing about environmental effects of AI technologies, especially the large amount of

energy used by Large Language Models (LLM) and its carbon footprint. This topic matters to me
because I want to work with LLMs in the future and I want these technologies to be more
environment friendly.

Thesis Statement: As deep learning models become more complex, they become more powerful
and accurate. However, they also become more energy demanding. Researchers must find a
way to balance model accuracy, size and energy usages.

Paragraph 2: Background Information

What general information do you want your reader to know/ understand about your topic? How
did the issue/ concern start?

AI is used in various field, such as healthcare, finance and self-driving cars. Most modern AI are
based on Deep learning models, that requires a lot of data and computing power. As the size
and complexity of these models grow, the energy required to train and operate them also
increases. According to Renée Cho, training just one large AI model can emit as much carbon as
five cars in their lifetime (Cho, 2023). As more models are trained and deployed, their carbon
footprint grows, raising a serious concern for the planet.

Paragraph 3: First issue/concern/point you want to make about your topic

Topic Sentence: Generative AI technologies like GPT-4 requires large amount of energy to
operate and even water for its cooling.

Introduce the author/ source of your evidence: Adam Zewe from MIT News clearly explains this.
Quote/paraphrase with in-text citation: He highlights that “large models tend to perform better
but require more computational power and energy” (Zewe, 2025).

Explanation/Examples/Analysis (2-4 sentences): AI developers often face a trade-off: improving

model accuracy might increases performance but also increases energy use. This energy
demand could lead to more pollution and harm the planet over time. This created a challenge:
how do we make strong AI systems without affecting the environment?

Paragraph 4: Second issue/concern/point you want to make about your topic

Topic Sentence: The rise of AI is contributing to global greenhouse gas emissions.

Introduce the author/ source of your evidence: Renee Cho discusses this in her article on
Columbia’s state of the Planet blog.

Quote/paraphrase with in-text citation: She says that training GPT-3 used around 1287 MWh of
electricity and emitted over 550 tons of CO2 gases( cho, 2023).

Explanation/Examples/Analysis (2-4 sentences) : This shows while AI technologies can be

helpful in solving many real-world problems, it is also a major source of environment pollution
and carbon emission. Cho also highlights that using model to generate response-known as
inference-can sometimes use more energy than the training process.

Paragraph 5: Third issue/concern/point you want to make about your topic

Topic Sentence: Not all communities are equally affected by AI large energy Usage.

Introduce the author/ source of your evidence: Adam zewe and AzoRobotics clearly highlights
this concern in their article.

Quote/paraphrase with in-text citation: Zewe (2025) reports that data centers usually placed in
areas with cheap electricity, increasing local emissions and resource usages in most cases.
AzoRobotics (2024) also highlights the ethical concern that communities in low-income regions
might have to face the environmental burden without getting the benefits of such technology.

Explanation/Examples/Analysis (2-4 sentence): The environmental impact of AI is not same

everywhere. Zewe (2025) explains that many data centers are built in places where electricity is
cheap, generally in rural parts of Aisa or US. These places often face more pollution and use a
lot of local energy and water resources. AzoRobotics(2024) says that Ai should be fair and
ethical. This means we need to make sure that some communities are not harmed more than
others. Everyone should share the benefits and cost of AI equally.

Paragraph 6: First solution/ step that should be taken

Topic Sentence: One way to reduce the energy used by AI is by making the models smaller and
faster.

Introduce the author/ source of your evidence: Tailin Liang talks about techniques to reduce the
models size and increase efficiency.

Quote/paraphrase with in-text citation: Liang et al. (2021) explains that pruning removes
unnecessary parts of the model, while quantization reduces the precision of numbers used in
AI, which helps to reduce energy use and increase performance.

Explanation/Examples/Analysis (2-4 sentences ): These techniques make AI models smaller and

faster, which helps to save energy. Moreover, the model accuracy also remains accurate enough
to be useful. If more AI developers uses these methods, AI Can become more powerful and
more sustainable at the same time.
Paragraph 7: Second solution/ step that should be taken
Topic Sentence: Developing ethical and sustainable Ai systems is required to minimize
environment pollution and balanced impact across all regions.

Introduce the author/ source of your evidence: AzoRobotics clearly highlights this in their article
about sustainable AI.
Quote/paraphrase with in-text citation: The article states that sustainable AI means designing
systems that reduce energy use and make sure no one group or area is harmed more than
others (AzoRobotics, 2024).
Explanation/Examples/Analysis (2-4 sentences): We need to use clean energy sources but also
address xthe unequal impact on rural communities. This means using clean energy sources like
solar or wind to power data centers and building AI in a way that does not harm poor
communities. Ethical design can make AI fairer and better for the environment. If we want a
better future, we must dbuild AI that cares for the people and the planet.
Paragraph 8: Conclusion
Revisit your thesis: As such models grow complex and the size increases, we need to balance
accuracy, model size and energy usage to protect the environment.

Review your main points: This paper has discussed energy consumption, carbon emissions, and
environmental inequality caused by AI. Moving Forward, it is vital for researchers, companies,
and policymakers to implement sustainable and ethical practices.

Why does this matter? Who will it impact? What will it mean to those involved?

If we don’t address these problems now, AI could do more harm than good in future. It will
affect our climate, our resources, and many communities around the world. If we don’t change
how we build and use AI, the damage will grow. But with better technology and fair thinking, we
can make AI better for everyone.

Thank them for their time and consideration. Provide your signature.
Thank you for taking your time and thinking about this important topic.
Sincerely,
Nishan Gadtaula