1 of 33

Basics of Using AI for Your Classes

Guy Wilson & Kirk Wilkins

1

2 of 33

Introduction

Topics covered in this session include:

Intro to AI
University policy regarding AI
What to include in your syllabus and assignment instructions
Update on AI text detection
Suggested tools
Teaching tips

Welcome. My name is Guy Wilson. I am an Instructional Technologist with Missouri Online. My colleague Kirk Wilkins, an Instructional Designer, and I will be covering a few basics of AI use for courses within the University of Missouri System. Angie Blaisdell, a Media Designer with Missouri Online, will be moderating chat.

A couple of caveats before we start. AI, especially the generative AI we will be covering, is part of a rapidly changing landscape of technologies, capabilities, and policies. While what we talk about today will be generally applicable, be aware that a lot can change in a matter of weeks. Also, we will be talking about policies or guidelines applicable to the entire UM System. As more policy and guidelines are rolled out, most of them are expected to come at the campus, school/college, or even department levels. Check with your campus teaching/learning organizations (T4LC, CTL, or CAFE), or your department to keep up with the latest policies and guidelines.

We will make a recording of today's event available along with handouts later this week.

3 of 33

What Do We Mean by AI?

What Is AI?

Today, we are talking about just one kind of AI, there are many, many fields within AI. It is a varied discipline. For the most part, we are talking about Generative AI, a sub-field of Natural Language Processing, but some tools may combine different types of AI for different purposes.

This slide is an attempt to show the variety of different uses of GAI. At the top, are general-purpose applications, most of which are also foundation models, that is other applications are built on top of them, but they also serve as applications themselves. GPT applications, for instance, include ChatGPT, Bing AI Chat, DALL-E 3, and much of the new Microsoft Copilot applications.

There are also stand-alone applications that have a specific purpose, such as Research Rabbit for academic research, or text-to-media applications that generate images, audio, or video output.

What we might call embedded applications for lack of a better term include tools like Microsoft Copilot or Adobe Firefly that live inside of other, often existing applications, such as Word or Photoshop.

Finally, we are seeing the rise of embodied applications, that is GenAI appearing in robots. These range from robot "dogs" like those from Boston Dynamics, to personal servants, like Neo, just now coming onto the market, and some applications in industrial or commercial robots. (See https://www.1x.tech/androids if we need to show examples.)

Though it does not fit readily into any of these categories, I would like to stop and say a word about Multimodal AI, which cuts across all of them. By "mode" we mean the form the input or output can take. Initially, for instance, ChatGPT was text-only. In October we saw new or improved capabilities rolled out that allow it to listen and speak, take image input, even identifying things in photographs and give instructions based on them, and respond with high-quality images through DALL-E 3.

4 of 33

What Do We Mean by AI?

What Is AI?

GPT picks tokens (words or fragments) that make sense in context.

Colors show how likely a word is. (Green most, Red least likely.)

The numbers show the position in the list. This varies by context.

How AI Works

5 of 33

University Policy on AI

6 of 33

University Policy

200.010 Standard of Conduct §C.1.c
Most policy devolved to campuses
Check with your department
Consider existing policies/procedures:

Academic integrity investigations
Software adoption
Simplified Tuition
Privacy, security, accessibility, equity

On June 29, the Board of Curators approved an addition to the Student Standard of Conduct regarding AI.

(i) and (ii) prohibit use of AI on assessments and assignments without permission from the instructor.
(iii) gives the instructor additional discretion to prohibit other uses in the class.
(iv) coordinates with other university policies
These are followed by an unnumbered provision that specifically allows use of grammar checking, autocorrect and autocomplete, and similar features of software such as Grammarly or even Word, unless the instructor specifically forbids it.

There is a slight change of wording to another part of the policy that emphasizes that instructors do have authority over what students can use and that these policies cover both cheating and plagiarism.

This will have implications for your syllabi and class policies.

Kirk will talk more about class policies, but I want to point out a possible area for confusion due to that last, unenumerated clause.

7 of 33

Gray Areas

Remember that AI can be used for almost any part of any assignment.
If you think a use of AI is not covered by the policy, and wish to prohibit, be explicit.
If you do not want students to use tools like Grammarly, or some parts of them, be specific.

Also Consider:

Privacy
Security
Student Copyright
Bias

Image by DALL-E 3

It is vital to keep in mind that AI can be used for any part of almost any online assignment, or soon will be. Design your class or assignment policies with that in mind. Things are no longer as simple as saying: thou shalt not copy. AI has very different implications from simple plagiarism or Googling things.

You may also wish to clarify specific actions that are prohibited if you think they are not sufficiently clear. For instance, you might want to prohibit students using an AI for brainstorming ideas for a paper or presentation.

Be aware that tools like Grammarly, Word, and Google Docs are going well beyond current ideas about grammar checking, spell checking, and autocomplete. They are incorporating more and more generative AI capabilities that will make them able to complete or generate student work in the same way that ChatGPT, Bard, or Claude can now.

There are a lot of areas that the University, either the System or the campuses, have yet to work through. These include privacy and security issues with unapproved applications (basically all GAI apps at this point), and student copyright issues. At this point, the best advice is not to put student work into an AI for evaluation. Also, beware that these AIs, even though they have safety rails, do reflect common biases and prejudices found on the internet. This may become an issue for you and your students depending on the type of assignment.

Also, remember to keep abreast of new policies from your campus or department.

8 of 33

Syllabus and Assignment Instructions

9 of 33

Goal for Students: TIP

Students should use generative AI with:

Transparency
Integrity
Permission

Framework developed by Christine Hanlon at UCF

10 of 33

What Is Your Policy?

Share with students what your policy is:

Red light (total ban)
Yellow light (limited and contextual use)
Green light (unlimited use with acknowledgement)

Image by Adobe Firefly

11 of 33

Why Is This Your Policy?

Image by Adobe Firefly

Explain to students why you have adopted your expectations.

Doing so will:

Focus on learning (outcomes)
Build goodwill and buy-in
Encourage compliance

12 of 33

Sharing Our Tentative Policies

In the chat, please share:

What your tentative policy is
Why you have adopted this policy

13 of 33

Communicating Policy

At Start of Semester:

Syllabus
Discussion

During Semester:

Assignment instructions
Assignment rubrics
Modeling
Discussion

14 of 33

Concerns with AI Policies

Diversity of policies among instructors
Potential need for flexibility and adaptation

15 of 33

AI Detection

16 of 33

Update on AI Detection

Conflicting views on worth
Concerns:

False Positives
Discrimination
False Negatives

Image by DALL-E 3

There is a debate on AI Detection. We normally only talk about AI writing or text detection but there are also tools for images, video, and audio. At this point, there are conflicting views on how accurate and usable they are. I'll talk here just about the text detectors.

What we do know is that they produce some false positives, the rates at which they do so vary. These are of concern for a number of reasons, not least of which are known discrimination against non-native writers, neurodivergent individuals, and native speakers of non-stanard English dialects. Even for those raised with completely standard English, the tendency of students to write college papers using more formal language may also trigger false positives. Unless students carefully track and log their work, they have very little way to prove that they actually wrote a paper that has been flagged.

You may have heard that Turnitin has a feature called AI Indicator. Currently, none of the campuses have activated this. If we did turn it on across the whole system, based on how much Turnitin is used, and on information on false positives from the company, we would expect to get between about 3250 and 4400 false positives across the System annually. That is between 0.74% and 1% of papers that would be marked as having some AI content.

Another concern is the rate of false negatives. That is what do the detectors miss?

17 of 33

False Negatives

Violet boxes indicate purely human content - scores should be 0%

Orange boxes indicate mixed human/AI content - scores should be be ~40-90%

All other is AI content - scores should be 100%

Overlapping bullets may hide others

This chart is based on some testing I did in the Spring and Summer of 2023 of four different detectors, each marked with a different color.

The X-axis show different papers. The ones that are in violet boxes are ones written purely by humans, while the ones in orange are mixed human-AI content. The others are all pure AI-generated content.

The Y-axis shows the percentage of content that each detector flagged as AI-created. Some of the detectors missed most of the content, on some papers, particularly those that used more advanced models or techniques designed to produce more fluent writing or fool detectors.

Different detectors, all using similar methodologies, come up with radically different results in about ¾'s of the papers. It also shows that instructors using them need decide how much suspected AI is necessary to appear before taking action. That, in turn, begs the question of what action to take. Unlike similarity detection, which gives direct evidence of pieces of text that match, AI detection is all probabilistic - the evidence is at best circumstantial. It is not clear how actionable it is.

You may have heard about watermarking AI content. While this may work for other kinds of content, it is so easy to defeat text watermarking that it is unlikely to ever be useful.

Finally, one last point: never, ever put a student paper into a chatbot, like ChatGPT, and ask if it was written by a human or an AI. The results are even worse than any of the AI detectors, and you can count on it to hallucinate reasons as to why it was written by an AI or human. One instructor at a Texas A&M campus did this and failed an entire class of seniors without other justification. He made the national news.

18 of 33

Suggested Tools

19 of 33

Live Demos

Copilot/Bing Chat Enterprise
ChatGPT
Perplexity
Research Rabbit

20 of 33

Teaching Tips

21 of 33

Areas to Consider

Law and ethics
Pedagogy
Prompt engineering

22 of 33

Law: FERPA

Federal Educational Rights and Privacy Act:

AI: DCL1 currently

No PII

Student use: mandatory or optional?
Teacher use: course design vs. assignment feedback

Image by Adobe Firefly

23 of 33

Ethics: TIP in Your Teaching Practice

Image by Adobe Firefly

Course design: Be transparent about when you have used generative AI.

Feedback: Obtain permission from students before using generative AI to evaluate their work.

24 of 33

Pedagogy

Potential responses to generative AI may include:

Process-centered teaching
Higher-order thinking
Reflection

25 of 33

Process-Centered Teaching

Image by Adobe Firefly

Product-centered teaching:

Focus on destination (final submission)

Summative assessments

Feedback only at end

Process-centered teaching:

Focus on journey

Formative assessments

Feedback throughout

26 of 33

Higher-Order Thinking

Bloom’s Taxonomy:

Generative AI may support lower stages.
This would give more time and space for higher stages.

Students can also analyze, evaluate, and synthesize output from generative AI.

Image from Vanderbilt University Center for Teaching

27 of 33

Reflection

Image by Adobe Firefly

What is it?
What is its value?
How could it play a role in learning with generative AI?

28 of 33

Reflecting on Our Teaching

In the chat, please share:

What: any changes you could make to your teaching
Why: the difference these changes would make

29 of 33

Prompt Engineering

Part of “AI literacy”
Student need for clear instruction and guidance

Image by Adobe Firefly

30 of 33

Prompt Engineering Strategies

Image by Adobe Firefly

Provide context.

Clarify rhetorical situation (purpose, audience, and genre).
Adopt personas.

Be specific.

Specify output expectations (formatting, etc.).

Keep going.

Use iterative/chained prompting.

31 of 33

Further Resources

32 of 33

Contact Information

Guy Wilson, Instructional Technologist IV

wilsong@umsystem.edu (meet by request)

Kirk Wilkins, Instructional Designer II

kwnzr@umsystem.edu (meet by request)

1 of 33

2 of 33

3 of 33

4 of 33

5 of 33

6 of 33

7 of 33

8 of 33

9 of 33

10 of 33

11 of 33

12 of 33

13 of 33

14 of 33

15 of 33

16 of 33

17 of 33

18 of 33

19 of 33

20 of 33

21 of 33

22 of 33

23 of 33

24 of 33

25 of 33

26 of 33

27 of 33

28 of 33

29 of 33

30 of 33

31 of 33

32 of 33

33 of 33