Rhodes Artificial Intelligence (AI): Home

Challenges:

• It's not always clear where AI is pulling information from. Some AI chatbots do cite sources to indicate what sites they've trained, but not all.
• AI often "confabulates" (generates false or misleading information), especially when it comes to academic references and citations. If AI hasn't trained on data that would answer your questions, it may synthesize that data in order to answer your question. Chatbots like ChatGPT are designed to provide text - the text answers it provides may not always be accurate or correct.
• Information AI has trained on could be outdated. For example, as of November 2024, ChatGPT's training data only goes to April of 2023. Though ChatGPT can engage with other current information online through its search tool, its training is dated.
• The datasets that chatbots train on could be based on incomplete or biased data. Because of this, implicit and systemic biases could be built into AI materials.
• AI can be used to create fake content, disinformation, or deepfake videos. Because of this, AI could be used maliciously.
• Ownership of content and copyright violations is a major challenge for AI. Depending on the terms of service, if you were to upload something into an AI platform, you could give that platform rights to or ownership of that content. If it trains on copyrighted material, it could inadvertently violate copyright and pass off copyrighted data as its own.

Possibilities

• AI chatbots are incredibly useful for brainstorming topics that you may want to write on. Again, consult with your instructor before using AI tools.
• Certain AI chatbots and tools are capable proofreaders.
• AI chatbots train on so much information that their vocabularies are enormous. This makes them capable of assisting you in developing searching strategies for finding books and articles.

Most AI chatbots are housed in data centers, even those that are operated on the cloud. In order for these data centers to function, they use up quite a bit of resources.

• The microchips that are used to power AI use rare earth elements which are often collected in environmentally destructive ways.
• Data centers produce electronic waste (mercury, lead, etc.)
• Cooling electrical components in data centers consumes a lot of water. AI-related infrastructure may soon consume six times more water than Denmark, according to one estimate.
• Many data centers burn fossil fuels to generate energy.
• One request through ChatGPT consumes ten times the electricity of one Google search.

Image created by Microsoft Copilot

The practice of formulating effective queries to elicit relevant responses from generative AI models is called Prompt Engineering. Knowing how to interact with and communicate with AI chatbots will yield more accurate responses and could reduce the amount of time spent asking AI questions, allowing users to conserve more resources. To build an effective prompt, be deliberate in what you're asking. It's okay to use multiple sentences to be clear about what you are asking. Consider the following guide developed by University of California San Diego:

Citing AI in MLA

As of December 2024, MLA does not recommend treating an AI tool as an author, and instead using the Title of Container element to specify the AI tool and the Publisher element to specify the company that made the tool. (Generative AI in MLA style)

Citing AI in APA

As of December 2024, APA style recommends citing the AI tool company as the author and AI tool as the title, with in-text citations and references adapted from the reference template for software in Section 10.10 of the APA Publication Manual. (Generative AI in APA style)

Citing AI in Chicago Style

As of December 2024, Chicago style does cite AI tool as author, with the following example: "ChatGPT is the author of the content, and the date is the date the text was generated. OpenAI (the organization that developed ChatGPT) is then listed as the publisher or sponsor of the content." (Generative AI in Chicago style)

Citing AI in CSE Style

Although there is no specific guidance offered by CSE, AI content is considered personal communication and/or non-recoverable data, and therefore should not be included as a reference in your reference list according to CSE style.