Skip to Main Content

Artificial Intelligence

What is AI?

AI vs. GenAI

According to author Gabriela Jhean, "Traditional (Artificial Intelligence)AI systems are designed to perform specific tasks or solve problems. They follow pre-defined rules and algorithms to accomplish these tasks. For example, a traditional AI system can be trained to recognize and classify images of trees and flowers based on certain features.'

'On the other hand, generative AI systems are not limited to specific tasks. They are trained on vast amounts of data and can generate new content based on the patterns they learn from that data. For example, a generative AI system can be trained on tree and flower images and generate new images resembling trees and flowers, even if they have never seen those specific images before." 1

Many different applications fall under the Artificial Intelligence umbrella term, including some types that are already integrated in every day life including speech-to-text, facial or fingerprint recognition, and text generation that tries to autofill in words as you type. 

family tree of artificial intelligence that shows the branching of different subtypes of AI as it interacts with fields like machine learning and data science, these terms are available in the AI definitions PDF above

This work2 is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License.

 

1 Jhean, Gabriela. (2024, May 21). AI vs. Generative AI: What's the Difference? MyCase. https://www.mycase.com/blog/ai/ai-vs-generative-ai/#:~:text=Unlike%20traditional%20AI%20systems%2C%20which,to%20create%20original%2C%20creative%20content.

2 Wheatley, A & Hervieux, S. (2020). The AI family tree [diagram]. The LibrAIry. https://thelibrairy.wordpress.com/2020/05/12/the-ai-family-tree/

 

AI Pitfalls

Our librarian asked ChatGPT to explain the downsides of using Artificial Intelligence for work in academics:

A screenshot of a conversation between a librarian and ChatGPT as the librarian asks the A.I. to explain three downsides for using its services in academic work and ChatGPT responds with the following reasons and explanations: potential for misinformation, lack of original analysis, and risk of academic integrity violations.

According to ChatGPT, "three key downsides to using me service for academic work:

  1. Potential for Misinformation – While I strive for accuracy, I may occasionally provide outdated or incorrect information, which can mislead research or assignments. Always verify sources.

  2. Lack of Original Analysis – I generate responses based on existing knowledge but do not produce truly novel insights or critical thinking, which are essential for academic work.

  3. Risk of Academic Integrity Violations – Using AI-generated content without proper citation or passing it off as original work could violate academic honesty policies, leading to serious consequences."

Potential for Misinformation

GenAI users should consider the following when engaging with AI for accurate and honest academic use:

  • Inaccurate Information and Hallucinations- Many AI models are trained and updated periodically, meaning that the information that the AI is pulling from might be from sources that are outdated. As GenAI content is created, it is shared into the same internet spaces that the GenAI is trained on, and thus making it harder to convince the AI that its information is wrong. Because people can easily feed false information to the GenAI, it will generate a response on the information it has. This will lead the AI to "hallucinate" that something is factual when it is not. Many GenAI, including ChatGPT, have a disclaimer for the results page that clarifies that AI can generate mistakes or incorrect information. 
  • Biases - The information fed into GenAI can also cause biases towards different populations and regions. Working with AI consistently can generate loops where the AI is not generating new content to users and instead will repeat the same assertion about the group or location. This bias is especially prevalent in image creation and text descriptions. Check out How AI reduces the word to stereotypes - Rest of World and ChatGPT Replicates Gender Bias in Recommendation Letters
  • Copyright - The United States Copyright Office has issued a report on the copyrightability of outputs generated by AI, reaffirming that works created entirely by AI are not copyrightable or claimable, work that includes both AI and human input may only have the human input copyrighted, and lastly that using Ai tool for the creative process including idea creation or editing does not render the entire work uncopyrightable.
  • Privacy - No users of AI should, under any circumstances, provide personal identifiable information into an AI site including name, date of birth, address, email, facial photos, etc. Companies creating AI platforms are training the models on text, images, audio, and video that may or may not have been obtained with the owner's permission. Clearview AI, a facial recognition startup, have been recently sued and settled out of court for violating subject's privacy rights for those pictured in photos that the company pulled from the internet into a database. This database was used to train AI that the company sells to businesses and government entities.

Lack of Original Analysis

The importance of assignments and projects that require critical analysis and reflection are two-fold: they are designed to make students think critically of and question the world around them. This is highly important to help people make decisions, communicate, and understand others. The assignments and projects are also designed to encourage students to explore topics through the lens of their own experiences and views, as well as expose the students to the views and experiences of others.

Risk of Academic Integrity Violations

As mentioned in the Fletcher Policy section above, the unauthorized use of GenAI will be considered "unsanctioned collaboration" and can result in action on the grounds of academic dishonesty. To avoid this, students are encouraged to contact their instructor if they have not outlined their instructions for the use or prohibition of AI use in the class syllabus or on assignments.