Introduction to Artificial Intelligence: For Faculty

This guide covers artificial intelligence (AI) in higher education, including background, integration in learning, ethics, and resources. AI as a technology and tools in education is quickly evolving, please check back for updates and developments. Along with AI technology, this guide is under construction and a work in progress. Many of the tools listed in this guide are under development and experimental and have not been evaluated by Iona University Libraries.  

"At the most basic level, a chatbot is a computer program that simulates and processes human conversation (either written or spoken), allowing humans to interact with digital devices as if they were communicating with a real person. Chatbots can be as simple as rudimentary programs that answer a simple query with a single-line response, or as sophisticated as digital assistants that learn and evolve to deliver increasing levels of personalization as they gather and process information."

Source: https://www.oracle.com/chatbots/what-is-a-chatbot/

Popular Chatbots

Some of these tools may require logins and accounts. 

ChatGPT - Chatbot from OpenAI with Free (GPT-3.5) and Plus (GPT-4) options.

Claude.ai - Chatbot from Anthropic.

Copilot - Chatbot from Microsoft. 

Gemini - Chatbot from Google.

"AI has become a catchall term for applications that perform complex tasks that once required human input such as communicating with customers online or playing chess." 

Source: https://www.oracle.com/artificial-intelligence/what-is-ai/

Elements of AI -  Online course from the University of Helsinki

Artificial Intelligence (AI) Terms: A to Z Glossary 

Useful Links

New ways to manage your data in ChatGPT (April 26, 2023).

Learn more about ChatGPT-4 from OpenAI.

Learn about the differences between ChatGPT-4 (available in the paid version) and ChatGPT-3.5 (available in the free version) from OpenAI.

ChatGPT's changelog provides updates on the latest changes to the tool.

ChatGPT Privacy Policies.

Data and Privacy

OpenAI (the company that designed ChatGPT) collects quite a bit of data from ChatGPT users.

• The privacy policy states that this data can be shared with third-party vendors, law enforcement, affiliates, and other users.
• This tool should not be used by children under 13 (data collection from children under 13 violates the United States’ Children’s Online Privacy Protection Rule - COPPA).
• The Terms of Use state that, “You must be at least 13 years old or the minimum age required in your country to consent to use the Services. If you are under 18 you must have your parent or legal guardian’s permission to use the Services” (OpenAI, 2023).
• While you can request to have your ChatGPT account deleted, the prompts and questions you input into ChatGPT cannot be deleted. If you were to ask ChatGPT about sensitive or controversial topics, this data cannot be removed.

TIP: Before using ChatGPT, please read over the privacy policy and terms of use and review how data is collected and shared as outlined in the policies. 

ChatGPT is not always trustworthy.

• ChatGPT was trained using a massive dataset of text written by humans that was pulled from the Internet.
• Responses can reflect the biases of the humans who wrote the text used in the training dataset.
• ChatGPT is not connected to the Internet, and as of April 2024, the data used to train ChatGPT 3.5 was collected up to January 2022.  The data used to train ChatGPT 4 is current up to December 2023.
• How should AI systems behave, and who should decide? (from OpenAI)

OpenAI. (2024). ChatGPT (April 15 version) [Large language model]. https://chat.openai.com/?model=gpt-4

Prompt:  What are the limitations of ChatGPT

Response:

ChatGPT, like any AI model, has several limitations that can impact its performance and utility. Here are some key limitations:

1. Knowledge Cut-Off Date: ChatGPT's training data includes information only up until a certain date (for this version, December 2023). Therefore, it cannot provide updates or knowledge about events, developments, or discoveries that occurred after this time.

2. Understanding and Context: While ChatGPT can generate responses based on patterns in data it has been trained on, it doesn't "understand" information in the human sense. It can generate plausible-sounding responses that may be incorrect or nonsensical, especially in complex or nuanced discussions.

3. Specificity and Detail: ChatGPT may struggle with very specific or technical questions outside common knowledge or highly specialized topics. It can provide general information but may not be able to handle deep technical details or very recent research in specialized fields.

4. Bias and Fairness: The model can inadvertently generate biased or offensive content. It reflects biases present in the training data, which can influence the tone and content of its responses.

5. Creativity and Innovation: While ChatGPT can generate creative content, it is limited to combining and reiterating patterns it has seen during training. It cannot come up with genuinely new ideas or creative outputs that are entirely novel.

6. Error Frequency and Correction: ChatGPT can make factual errors and might not always correct them, especially if it's not prompted to review or reconsider its responses.

7. Interaction Limitations: The model operates within predefined interaction guidelines and ethical considerations. It will not engage in generating harmful content or participating in activities that breach privacy or ethical standards.

These limitations mean that while ChatGPT can be a powerful tool for generating text-based content and automating interactions, it requires careful management and oversight, especially in sensitive or critical applications.

Content used and modified with permission from Torrey Trust at University of Massachusetts Amherst. Retrieved January 19, 2023. https://docs.google.com/presentation/d/1Vo9w4ftPx-rizdWyaYoB-pQ3DzK1n325OgDgXsnt0X0/edit?usp=sharing

Never enter private or confidential information into ChatGPT and similar tools. By default, ChatGPT uses your data to improve its performance and capabilities. However, you have the option to prevent ChatGPT from using your data for this purpose.

What Does Training Mean?

Learning from Data:
Just like a human learns from experience, AI learns from large amounts of data, including text from books, websites, articles, and more.
During training, AI analyzes this data to understand patterns, language structures, and various topics.

Improving Performance:

Training helps the AI get better at understanding and generating human-like text.
It learns to respond accurately and appropriately to different types of questions and conversations.

Why Training Uses Your Data:

Sometimes, AI systems use real interactions (like your chats) to fine-tune and improve their responses. This makes the AI more useful, relevant, and accurate for future interactions.

Prevent ChatGPT from Training with Your Data: Why Turning Off Training with your Data is important

If you turn off training, your interactions will not be used to improve the AI. This helps maintain your privacy and control over your data.

Privacy: Ensuring your personal information and conversations are not used for training can protect your privacy.

Control: Having the option to turn off training gives you control over how your data is used.

How to View Your ChatGPT Data (Memories)
To request ChatGPT to list your stored memories, follow these steps:

Initiate a Conversation by starting a new chat session with ChatGPT and ask ChatGPT to list all stored memories.

For example:

• "Can you list all the memories you have stored for me?"
• "Please show me all the data you have remembered from our previous sessions."
• "What information have you stored about me?"

ChatGPT will then provide you with a list of all the stored memories or data from your previous interactions for you to review.

How to Disable Training: To turn off ChatGPT training on your data, follow these steps provided by OpenAI:

• Access Your Settings: Go to the settings in the ChatGPT interface. This is usually found by clicking on your profile picture or name, typically located in the upper right corner of the screen.
• Manage Data Controls: Look for a section or option labeled "Data Controls" or something similar.
• Disable Data Sharing: In the Data Controls section, find the option to disable data sharing. Toggle this option to turn off the sharing of your data for training purposes.
• Confirm Changes: Ensure you save or confirm any changes you make in the settings to apply your new preferences.

You can also make a privacy request to OpenAI by visiting OpenAI Privacy and clicking the button labeled ‘Make a Privacy Request’. You will be presented with four options:

• Download my data: Request a copy of your data.
• Do not train on my content: Ask OpenAI to stop training on your content.
• Delete my OpenAI account: Request to delete your account (only use this option if you wish to no longer use ChatGPT).
• OpenAI Personal Data Removal Request: Request the removal of your personal data from AI model outputs.

Another option is to use the feature called “temporary chat.” At the top of the page, click on the menu that says “ChatGPT,” and then click the option that says “temporary chat.” Then your chat won’t appear in your history and ChatGPT won’t save anything from your conversation.

You can do the similar in other tools. In Perplexity’s settings, the switch is called “AI data retention.” In Google’s Gemini, click a button in the lower left called “activity.” At the top of the page that comes up, you can click the menu that says “turn off.” Claude doesn’t use your inputs for training unless you opt in.