Skip to main content Skip to secondary navigation

Defining AI and chatbots

Main content start

With this module, we aim to provide you with an introductory explanation of generative AI tools. We will explain key concepts and terms, as well as how AI tools are developed.

Key points from the previous module

Warming up to AI tools through play

  • Reflect on your own feelings about AI and why you feel that way.
  • Humor can be a useful way to increase curiosity and motivation to learn and engage.
Go to the previous module

Outcomes for this module

In this module, we want you to form a basic understanding of AI tools and feel prepared to begin using a chatbot in subsequent modules.

After completing this module, you should be able to:

  • Define common terms and concepts in AI.
  • Summarize how generative AI tools work.

AI already is in our daily lives

You likely already have more experience using AI tools than you know. They have become part of our daily lives for quite some time. We can describe any technology that predicts an outcome based on large sets of data as a form of artificial intelligence. For example, a streaming service platform recommending a new movie to you makes a prediction based on your past viewing history and patterns among wider groups of viewers. A map tool gives you navigational directions that predict the best route based on data from other drivers.

Begin with the warm-up question, “What AI-powered tools might you have already used?” and respond to the poll below.

Embed Code
Embed Code

Common AI terms

Artificial Intelligence (AI)—"The capacity of computers or other machines to exhibit or simulate intelligent behavior." (Oxford English Dictionary, n.d.)

Generative AI—A type of AI technology that generates content such as text, images, audio, and video. Also sometimes referred to as a generator.

Model—An AI software program that has been trained on datasets to perform a specific task.

Large language model (LLM)—A complex model trained on vast amounts of data that generates language that resembles human-generated language. GPT, PaLM, Jurassic, and Claude are examples of LLMs.

Chatbot—A computer program that uses an LLM to simulate a conversation with human users, typically through typed text in a software application.

Machine learning—A technique by which a computer can learn without being directly programmed with rules.

Deep learning—A subset of machine learning inspired by how biological brains are structured. Deep learning uses multiple layers of machine learning for progressively more sophisticated outputs.

Training—This refers to machine learning and deep learning processes used to develop a useful model.

Training data—Labeled data used in the training process to "teach" an AI model or algorithm to make a decision. For example, with an AI model for self-driving vehicles, training data may include images and videos in which traffic signs, pedestrians, bicyclists, vehicles, and so on are labeled.

Algorithm—A set of instructions or rules for performing a computation. Developers typically design algorithms used in AI to progressively iterate themselves, which we can consider a form of machine learning.

Alignment—How well an AI model aligns with humans' intended goals or ethical principles. An AI model is considered misaligned if it advances some objectives but not those intended by the human developers (Russell, et al., 2020). 

Beta test—In software development, a beta test is an opportunity for real users to use a product before a general release so that the developers can refine the product.

Prompt—Instructions entered by users to direct an AI generator to generate an output or complete a task.

How generative AI chatbots are developed

First, developers start with a set of labeled data. Then, they select a machine learning model to analyze the data and make predictions or identify patterns. Next, human software developers train the model by updating the data, adjusting the model parameters, or reinforcing the algorithm until it consistently produces the desired outputs. During this process, the algorithm within the model continuously updates itself.

In some cases, the training may use other methods that do not rely on direct human intervention, such as pattern recognition or programmed incentives (Brown, 2021). After training, the developers validate the model by inputting new data and testing if it can perform reliably. Finally, the developers may create different software applications that apply the AI model in a more usable way.

The development of AI chatbots, which are powered by sophisticated LLMs, usually requires substantial investment from large organizations or companies. Often, these companies release AI chatbots for free to engage users in a beta test where user data is gathered to further enhance the model.

Human-generated and AI-generated language

We take the broad view that language embodies a fundamental aspect of human life profoundly intertwined with every kind of human endeavor. AI tools that generate language thus have profound implications for our understanding of the meaning and purpose of language.

In our view, human language is more than a means of conveying information. The way humans generate language inherently includes self-exploration, self-expression, and relating to others in a way that AI tools do not (Diogenes 2023). AI chatbots can benefit us by increasing efficiency, generating information, and reducing the drudgery of certain reading and writing tasks. But focusing too narrowly on language as solely a means of transmitting information might lead to the use of chatbots as only an informational efficiency tool. Some might come to hold the belief that because the output of AI language generators resembles human-generated language, we lose nothing if the AI language conveys information clearly. This might be useful in some situations in which we make information or efficiency the priority. However, this could lead to students using language generated by chatbots as their own, or to instructors doubting the usefulness of reading and writing in the learning process.

We believe that in most educational contexts the human process of generating language has unique benefits and value for those engaging with language. Self-exploration, expression, and connection to each other remain critical to learning. We believe that students who hold this view might use chatbots to enhance, not replace, their voices, and that instructors will use reading and writing as ways of deepening the learning process, not as just a means of demonstrating retention of content.

Assess and reinforce your learning

We offer this activity for you to self-assess and reflect on what you learned in this module.

Stanford affiliates

  • Go to the Stanford-only version of this activity
  • Use your Stanford-provided Google account to respond.
  • You have the option of receiving an email summary of your responses
  • After submitting your responses, you will have the option to view the anonymized responses of other Stanford community members by clicking Show previous responses.

Non-Stanford users

  • Complete the activity embedded below.
  • You have the option of receiving an email summary of your responses.
  • Your responses will only be seen by the creators of these modules.
Embed Code

Learn more

Works cited

Brown, S. (2021, April 21). Machine learning explained. MIT Sloan. https://mitsloan.mit.edu/ideas-made-to-matter/machine-learning-explained

Diogenes, M. (2023, June 7). "From the Bridge: Chatbot Confidential: Machine Magic and the Power of Language". Stanford Teaching Writing. https://teachingwriting.stanford.edu/news/bridge-chatbot-confidential-machine-magic-and-power-language

Oxford English Dictionary. (n.d.). Retrieved July 19, 2023, from https://www.oed.com/?tl=true

Russell, S. J., & Norvig, P. (2020). Artificial intelligence: A modern approach (4th ed.). Pearson. pp. 31–34. ISBN 978-1-292-40113-3. OCLC 1303900751. 
 

Preview of the next module

Exploring the pedagogical uses of AI chatbots

An exploration of the capabilities and use cases for AI chatbots in teaching and learning contexts.

Go to the next module

Learning together with others can deepen the learning experience. We encourage you to organize your colleagues to complete these modules together or facilitate a workshop using our Do-it-yourself Workshop Kits on AI in education. Consider how you might adapt, remix, or enhance these resources for your needs. 

If you have any questions, contact us at TeachingCommons@stanford.edu. This guide is licensed under Creative Commons BY-NC-SA 4.0 (attribution, non-commercial, share-alike) and should be attributed to Stanford Teaching Commons.