Generative AI and Information Literacy

As the use of generative AI has increased, the effects of this technology on the environment have grown. Computer scientist Kate Saenko explains, "The more powerful the AI, the more energy it takes" (2024). The full extent of generative AI's energy consumption is not completely known yet. AI companies have kept some of this information secret and researchers are still investigating environmental impacts. However, it is clear that generative AI is likely to substantially affect the environment, as "generative AI requires powerful servers, and the worry is that all that computing power could make data centers' energy consumption and carbon footprint balloon" (Calma, 2023).

How exactly does generative AI impact the environment?

• Energy consumption: A huge amount of computing power is needed to create and train generative AI models and to generate its outputs. According to research, "It's estimated that a search driven by generative AI uses four to five times the energy of a conventional web search. Within years, large AI systems are likely to need as much energy as entire nations" (Crawford, 2024). In other words, typing a prompt into ChatGPT results in much more energy use than typing a search query into Google.
• Water use: The data centers that house generative AI servers consume an enormous amount of energy and resources. Much of this resource consumption consists of the water needed to provide cooling for computer servers. Shaolei Ren, a professor of engineering at UC Riverside, "estimates that a person who engages in a session of questions and answers with GPT-3...drives the consumption of a half-liter of fresh water" (Berreby, 2024). Ren and colleagues suggest that "globally, the demand for water for AI could be half that of the United Kingdom by 2027" (Crawford, 2024).
• Creating demand for more energy resources: The computing power needed for generative AI has driven tech companies to seek out more energy sources. This has included proposing nuclear energy as a solution (Crawford, 2024) or increasing the consumption of natural gas (Kimball, 2024). These energy demands would further impact an environment already strained by climate change.

Researchers have shown that it is possible to reduce the energy costs of generative AI by using more renewable energy, implementing sustainable construction of data centers, and scheduling computation during certain times of the day (Saenko, 2024). These practices would require transparency and commitment from tech companies and advocacy from users and policymakers.

In the conversation surrounding AI, the text produced by chatbots is often presented as the result of machine intelligence only. However, journalists have shown that AI text is not only the work of machines. Instead, the work of many human laborers is essential to the text generated by ChatGPT and other chatbots. According to an investigative report, "Behind even the most impressive AI systems are people — huge numbers of people labeling data to train it and clarifying data when it gets confused" (Dzieza, 2023).

How do humans contribute to the work of generative AI?

• Annotation: People are given data gathered from the Internet and label this data: for example, they assign emotions to people's voices in video calls or on social media posts, or they categorize images of items such as clothing or food. These labels are used to train AI to recognize and assign categories to data.
• Reinforcement learning from human feedback (RLHF): People "converse" with an AI chatbot and rate its responses for qualities such as authenticity or helpfulness. Engineers use these ratings to train the AI model to sound more "humanlike."
• Detecting toxic content: Similar to annotation, people identify and label "toxic" content from the Internet (including violent, disturbing, and harmful content). This is used to train the AI model to exclude such content from its generated text.

The human labor used to train generative AI models is often outsourced to underpaid workers in the Global South. For instance, workers in Kenya were paid less than $2 an hour to label disturbing toxic content (Perrigo, 2023). Some academics refer to these practices as "digital neocolonialism": Western tech companies exploit the labor and natural resources (for example, minerals used in computer hardware) of poor nations in the Global South, further perpetuating the legacy of colonialism (Browne, 2023).

On the Environmental Impacts of Generative AI

Berreby, David. "As Use of AI Soars, So Does the Energy and Water it Requires." Yale Environment 360, February 6, 2024. https://e360.yale.edu/features/artificial-intelligence-climate-energy-emissions.

Calma, Justine. "The Environmental Impact of the AI Revolution is Starting to Come into Focus." The Verge, October 10, 2023. https://www.theverge.com/2023/10/10/23911059/ai-climate-impact-google-openai-chatgpt-energy.

Crawford, Kate. "Generative AI's environmental costs are soaring -- and mostly secret." Nature, February 20, 2024. https://www.nature.com/articles/d41586-024-00478-x.

Saenko, Kate. "A Computer Scientist Breaks Down Generative AI's Hefty Carbon Footprint." Scientific American, May 25, 2023. https://www.scientificamerican.com/article/a-computer-scientist-breaks-down-generative-ais-hefty-carbon-footprint/.

On Human Labor and Generative AI

Browne, Grace. "AI is Steeped in Big Tech's 'Digital Colonialism.'" Wired, May 25, 2023. https://www.wired.com/story/abeba-birhane-ai-datasets/.

Dzieza, Josh. "AI is a Lot of Work." The Verge, June 20, 2023. https://www.theverge.com/features/23764584/ai-artificial-intelligence-data-notation-labor-scale-surge-remotasks-openai-chatbots.

Perrigo, Billy. "OpenAI Used Kenyan Workers on Less Than $2 Per Hour to Make ChatGPT Less Toxic." TIME, January 18, 2023. https://time.com/6247678/openai-chatgpt-kenya-workers/.

• Does learning about the environmental impacts of generative AI affect the way you will use it in the future? Why or why not?
• How does knowing about the way human labor is used to develop generative AI affect the way you think about or evaluate the text produced by AI? 
• Are you surprised in learning about generative AI's impacts on the environment and use of human labor? If so, how does this information challenge or alter your perceptions of generative AI?