Navigating Generative AI in Higher Education

Every algorithm, including machine learning tools, is influenced by the biases of its creators. Big tech companies are currently the ones taking the lead in LLMs and their own biases are being replicated in the algorithms' response. Algorithmic bias is not limited to AI tools but as these tools are integrated into more everyday systems, such as iPhones and Microsoft Office, users need to think about what viewpoints are missing from the conversation when technology creators remain disproportionately white and male. . 

Another consideration for users are the datasets used to train LLMs. Most of these are English-language based which means that the majority of generative AI tools lack a multilingual, multicultural perspective. It is important to be aware of this limitation and to seek out other sources that can provide diverse perspectives. 

In an effort to curb bias and harmful responses generated by AI tools, OpenAI (which powers Microsoft's CoPilot) and Google have put in place guardrails to protect users from disturbing outputs. While these have eliminated some problematic responses, there have been unintended consequences. For example, if you ask Google's Gemini where to write a story about a girl from the Gaza region or to rewrite the Gettysburg Address in the style of a modern-day teenager, it will be unable to do so as the current guardrails block both responses. In the meantime, users have found that the guardrails of CoPilot and ChatGPT can be easily circumvented. While the guardrails were created with good intentions, how they work is unclear, sometimes even to the developers. As AI continues to evolve and develop, the question of who should be deciding what content constitutes "disturbing" or "valuable" information. 

As of February 16, 2025, OpenAI is working to re-train their chatbots to "uncensor" content. This would eliminate content moderation of Chat-GPT but could increase responses that are biased, harmful, and incorrect. 

In 2023, the U.S. Copyright Office began an initiative "to examine the impact of generative Artificial Intelligence (AI) on copyright law and policy." This work is ongoing and until it is complete, the question of what constitutes fair use in the age of AI is unclear. As of July 31, 2024, the U.S. Copyright Office has recommended "that Congress enact a new federal law that protects all individuals from the knowing distribution of unauthorized digital replicas." This would protect against things like "deep fakes", which are videos or audio created to show an individual or group doing or saying something that did not actually happen. It would also address manipulation of digital content. The "No FAKES Act of 2024" was quickly introduced in the Senate and is moving through the legislative process. 

Legal actions against AI started with image creation software that steal images of protected artwork to create new images. AI companies have been accused of purchasing research datasets which allow them access to materials they otherwise could not access. The New York Times filed a lawsuit against OpenAI for scraping their database for training purposes, leading to the question of how they gained access to these datasets in the first place. LLMs are incorporating and using the words of authors and creators without permission and without crediting their work. Not knowing where ChatGPT is getting its information not only hurts the credibility of the information you gather but also erases the original human creator. 

In July 2024, Wiley and Taylor & Francis, publishers of academic work, both announced they had signed agreements to allow access of academic work to tech companies. This is a major concern for scholars who may not want their work used to train AI models. But agreements signed with these publishers may mean academics have no say in how their work is used after publication. 

The environmental impact of generative AI is substantial. Not only do these tools have a large carbon footprint, they require enormous amounts of water. One assessment, reported by Nature in February 2024, states that ChatGPT alone "is already consuming the energy of 33,000 homes." A report from October 2023 stated  that "the global AI demand may be accountable for 4.2 – 6.6 billion cubic meters of water withdrawal in 2027, which is more than the total annual water withdrawal of 4 – 6 Denmark or half of the United Kingdom." 

As LLMs grow, so does their carbon footprint. The full extent of how much AI systems are contributing to climate change are unknown - largely because the companies creating these tools are not tracking the environmental impact. But what is known so far is concerning. The following articles outline the environmental risks associated with wide adoption of AI tools. 

• AI Programs Consume Large Volumes of Scarce Water
• As Use of AI Soars, So Does the Energy and Water it Requires
• Tackling AI's Climate Change Problem
• AI's Growing Carbon Footprint
• The Environmental Impacts of AI Primer
• How AI Tech Companies are Obscuring AI's Real Carbon Footprint
• Constellation Energy to Restart Three Mile Island Nuclear Plant, Sell Power to Microsoft for AI

It should shock no one to hear that the internet is a toxic place, filled with vitriol and racism. And this is ChatGPT's learning ground. The first generations of the LLM parroted back this problematic material in its responses. To fix it took human labor, specifically Kenyan laborers earning $2 per hour, whose job it was to look at the text from the worst part of the internet and label examples of violence, hate speech, and sexual abuse. And ChatGPT is not alone. Data enrichment professionals from across the world have to work through lines of data to make these systems safe for public consumption. "AI often relies on hidden human labor in the Global South that can often be damaging and exploitative. These invisible workers remain on the margins even as their work contributes to billion-dollar industries." 

Misinformation can be spread in multiple ways through the use of generative AI tools.

First, AI chatbots want to be seen as useful to users. This sometimes leads to "hallucinations" or false information being presented as facts in response to a user prompt. This is especially problematic with tools that provide no citations, such as ChatGPT and Anthropic's Claude. For tools that do provide citations, you still must verify the source is real and the information provided is accurate. 

Second, the training data used for LLMs has not been fact checked or verified. Any inaccuracies, falsehoods, or conspiracy theories found on the internet have been used to train these tools, which means they may provide content that contains these same inaccuracies, falsehoods, or conspiracy theories. Examples of this have been widely documented and so far, the companies that host these tools have not adequately adapted their data to account for this problem. The only solution has been to have humans methodically go through the training data to remove information that is incorrect but this is an ethical dilemma as well (see "Labor Practices" above.) 

Finally, there is a rise in disinformation creation using generative AI tools to spread malicious content, particularly through social media accounts. Deep-fake videos have been spread to undermine political candidates in multiple elections across the globe, including the current 2024 United States election cycle. As this technology develops, individuals may find it more difficult to decipher truth from fiction. 

Using AI tools such as ChatGPT to complete homework assignments is considered plagiarism and violates OWU's Academic Honesty Statement. 

If a faculty member has given a student permission to use AI tools to complete an assignment, students should make sure to cite how and when AI was used. Proper attribution can ensure a student is not taking credit for work created artificially. Students should never use AI tools academically unless explicitly told to do so by their instructor and then only in the exact context given by that specific instructor. 

MLA  Style Citations for AI 

APA Style Citations for AI

Chicago Style Citations for AI

Before you use generative AI tools, you may want to consider the potential privacy risks. Many tools use user-interactions with Chatbots for training purposes. While companies like OpenAI and Google state that all identifying information is removed first, there is still a risk associated with offering personal information or details to AI tools. CNN Business published an article that highlights some of the unique aspects of generative AI and privacy in March 2023.

As with any technology, it is advisable to familiarize yourself with the privacy policy and user agreement of the product you are using prior to sign-up. It is a good rule of thumb not to share any information with an AI chatbot that you would not want to be freely available to the public.