Portland state University
CS 510 – Explorations in data science
Project plan
With the ever-expanding use of machine learning models, concerns such as ethical and legal issues are vastly discussed in many places. But should we be apprehensive about the power consumption of the ML and pollution that it makes? This paper’s main topic is around the carbon impact of the ML model training.
Initially, I will read the articles supporting that “the carbon impact of AI” is considerable. Afterward, I will research the carbon impact of other industries that are impacting our planet at the moment. Comparing the result of these two allows me to conclude whether the pollution made by AI is worrisome or not?
I would not create any ML code to measure the time and power consumption of it. Instead, codes from other articles and their report on the power usage will be used. Subsequently, those numbers will be applied to have a rough estimation of how much energy will be consumed to train a sophisticated ML model.
This team consists of only one person which would be Damon Aliomrany. Therefore, all the tasks will be done by me.
- Research on the supporting articles.
- Collect data from existing codes to estimate the power consumption and pollution of ML models.
- Prepare a paper about the result of this research on how bad is the carbon impact of AI.
- Research on other industries and compare the pollution they make to the AI field’s pollution.
- Reveal the final observation in the final paper and presentation.