Introduction to Generative AI in Assessments

There is a huge assortment of GenAI tools available to the public. Below is a non-exhaustive list of some of them, selected to show the variability of tools and usages available and of the most commonly used ones in education.

Written Text

The GPT model focuses on language and can be used to search for information and generate text (e.g., explanations, examples, short stories, poems, outlines, essays, codes for computer programming, lesson plans, tables comparing two things, etc.). All interactions take place in a written format. Several languages can be used. As mentioned on the previous page, the GPT model can be accessed through two common interfaces:

• The OpenAI website requires registration. It provides access to GPT-3.5 free of charge, and GPT-4.0 requires a subscription.
• The Microsoft Edge browser now embeds the Bing interface to GPT-4.0. It is freely available. To access it, download and open Edge and click on the Bing icon in the top right corner of the screen (it’s a fancy B icon). Bing is available in three modes: Creative, when you want it to create unique, new materials (this is the setting recommended for this course); Precise, when you need it to be more accurate and want it to stay matter of fact (it will often return the safe response to the same prompt), or Balanced, for something in between. If you want to learn more about these settings, consult this short Ars Technica article.
  
• Bard is Google’s version of a conversational GenAI chatbot (though not currently supported in Canada)
• You.com is another text-based search engine that functions much like ChatGPT.

Here is a short [3 min] tutorial on how to use ChatGPT.

Here is a short [3 min] tutorial on how to use Bing in Creative Mode.

• The Microsoft Edge browser now embeds the Bing in Creative Mode interface to GPT-4.0. It is freely available. To access it, download and open Edge and click on the Bing icon in the top right corner of the screen (it’s a fancy B icon). To get it to draw an image, enter a prompt that asks the GenAI to create an image.
• Firefly is an Adobe product that generates images based on a text prompt. There are filters that allow you to quickly and intuitively modify the initial image (e.g., re-creating the same image in a different style). It's currently free to use, but requires you to create an account and to sign in. That's because each user gets a set number of free images each month.
  
• Padlet is a visual digital discussion or message board. In addition to posting text, their own images, and links, users can post a GenAI image on any Padlet board. Educators must have an account (including the free account that limits the number of boards to three). Students do not need to have an account. Once they access a board (provided by their teachers), they start a new post (click on the + symbol) and then click on the (…) [three dots – more attachment types] . From there, they can click on the option “I can’t draw” and enter a prompt for their GenAI image. They will be provided a few options and can post that image as their post. (You will get a chance to experiment with this in the first discussion board).
• Stable Diffusion is an open source image generator. It does not require any registration. However, it is very slow (you have to wait in a queue to see the results of your text prompt for a few minutes).
• DALL-E is put out by OpenAI (the creators of ChatGPT) and uses the GPT-3.0 model to create images from text. It uses a credit system to limit the number of free images that users can create each month.
• Leonardo.AI is a tool that requires registration (and you can expect to be put on a waitlist when you register). Many reviews say that its capabilities are superior to other GenAI image services. It is a paid service, though there is a free plan with limited access (150 images per day).

Videos

You don’t need to go our with fancy video and audio equipment to create short, artificially generated videos anymore. You can tell the GenAI what you want via a text prompt, or give it the URL to your website, and ask it to create a short audio-visual summary. All platforms below require a registration (which solicits your email address) and offer some (limited) free services.

• D-ID is the digital people platform. You enter text, choose the presenter (the way they look and their voice, including their accent and speed of speech), enter what you want them to say in a text prompt, and the platform returns a person of your choice “reading” that text. See the attached example.
• RunWay is another platform that creates short videos from text.
• An example is provided, using the prompt: “A robot walks among a group of human university students and talks with them.”
  
  
• Fliki.AI is a tool that allows you to enter a text-based prompt and the website returns a video. You can feed it a URL from your blog site, or upload your Power Point file, and ask the system to create a short “trailer” video to promote it.

Power Point Slides

Did you know that the GPT model can also create your slide deck? You give it a prompt and it creates all of the slides, complete with a title slides, outline, images on each slide (with references) and including the script of what the presenter should say.

• SlideGPT. Is free to experiment with. Once the presentation has been generated you can view the whole thing. To download it as a Power Point file, you will need to pay a small fee.

Website Design

Text prompts are modifying how people interact with apps. Rather than click on each characteristic of the product and modify it to fit their needs, users interact with a chat box and tell the software how to modify the output using conversational language.

• Studio.AI is a GenAI tool that allows users to create website and modify the design using conversational prompts rather than by modifying each feature using direct buttons and tools. (e.g., “Wrap the image in a pink box”)