I, like many teachers, am struggling to understand the significance of, and usefulness for, newer forms of AI, such as ChatGPT. Two big trends I have noticed are: 1) the rush to include large language model abilities like ChatGPT in a variety of software products, and 2) the emerging need to develop skills in manipulating chat-style AI to get desired results.
AI in Notion
Notion, a general purpose note-taking and database web application, has a free education subscription price. It was one of the first services that offered an AI agent to its note-taking abilities. It has begun to charge for usage of the AI facility, which I think is another trend to watch for. Other services like Craft, a note-taking/writing web app, and Taskade, a web-based task manager, have added AI as well.
One place you can use AI is when you add a new page in Notion. Here, it offers to create a number of types of written documents based on your prompt or you can ask it to write something that doesn’t fit the given types.
AI Assist menu in Notion
I have used Notion for a few years as a web clipper, primarily, but it has many useful functions. To get the full benefit of all it can do does take some time and effort in learning. If you are interesting in learning how to use Notion there are a few Notion gurus who have good how-tos including Marie Poulin and Thomas Frank.
Chat prompting skills
Besides AI popping up in all sorts of services, another emerging trend is the skill of developing methods to get high-quality results from a chatbot like ChatGPT. A site on this topic called Awesome Chat Prompts provides some suggestions. This is also a good site to see all the possible uses of a large language model AI.
Related to prompts is the ability to supplement the data the AI uses with your own data. For example, my original interest in this sort of service was a desire to create drafts of quiz and test questions, similar to what you might see in commercial textbook instructor’s manuals, for my OER textbook. The data in ChatGPT from the internet doesn’t have the specific terms and ideas from my textbook, so I would need to supplement it with those unique terms and definitions. Feeding these models your own data, beyond what a prompt contains, is not easy yet. There are some experiments going on now using tools like Llamaindex, but it is early days for this sort of manipulation.
There is so much activity around this technology that I know it will be awhile before we see the true impact of it, but these are two trends to watch for as we try to understand and use it.
Update-April 26
Here is a post from Richard Byrne in which he compiles a collection of 15 AI tools he has written about. He is a K-12 blogger but much of the tech he discusses can be useful for college faculty as well.
Richard Byrne, a K-12 edtech blogger, posted this about three tools that have emerged to detect AI writing. While none of the tools are foolproof, and detection is not necessarily the most interesting aspect of the emerging generalized AI technology, I know some faculty would like to know what they can do to try to determine if a text was generated by AI.
Much is currently being written about AI (artificial intelligence), particularly ChatGTP (https://chat.openai.com/chat), a chatbot created by OpenAI. So much has been written, in fact, that I am not sure what more there is to say. For that reason, I will only make a few observations and provide a number of links to explore. (For more background on the terminology and categories of AI, here is a web page from IBM – https://www.ibm.com/cloud/learn/what-is-artificial-intelligence.)
What is ChatGPT?
ChatGTP is not the first or only use for AI. It is the first readily available “strong artificial intelligence” resource. It takes the form of a general AI, sucking in a snapshot of the information on the Internet and synthesizing it to a user’s specification. The user can ask almost any question, and ChatGPT will attempt to provide an answer, similar to chatbots like Amazon’s Alexa. Unlike Alexa, it is aware of the context of the conversation with the user, so the user can refine the results by giving further instructions in the chat. Compared to a Google search, it can produce a wider range of results. While Google can return a list of URLs or a brief summary of information, ChatGPT results can take nearly any form, including longer written works, freeform lists of ideas, and even computer programs. The potential time-saving is enormous, but it has some drawbacks. Much like Google, the process it uses to return results is opaque, and it is likely to stay that way as it is created by a for-profit company. Its results, based on a snapshot of the Internet and ChatGPT’s transformation process, can be inaccurate as well.
Other uses and forms
While the promise of an all-purpose Internet oracle or a free ghostwriter seems tantalizing, other uses for AI are also important. More specific uses, called “weak AI”, have been around for quite some time, ranging from AI that enhances photos to grammar checkers to audio transcription. I am particularly interested in those uses that save me time and typing by moving information from audio to text and back again. Newer specialized uses of “weak AI” still deserve attention as strong AI continues to be developed.
Besides ChatGPT itself, development of other tools that use strong AI are being developed. A starting point for learning about these is OpenAI’s page of examples- https://beta.openai.com/examples/. Also, a Chronicle of Higher Education article on ChatGPT (https://www.chronicle.com/article/ai-and-the-future-of-undergraduate-writing) mentions two other sites with newer AI tools, one to assist in research (https://elicit.org) and the other for creation of more AI tools (https://fermat.ws). In addition, Notion, which I have used off-and-on for note-taking and other tasks, has recently incorporated AI into its system. Below is a screenshot of an AI-generated post and a menu of AI options from Notion.
Screenshot of Notion’s AI Assist options
Here is a link to Notion’s guide to its AI facility, which has some useful information about AI in general – https://www.notion.so/help/guides/using-notion-ai. Notion’s AI Assist is in beta at the moment, so it is not readily available yet. However, Notion itself is free to college faculty (https://www.notion.so/product) and you can get on a waiting list for AI Assist. I like Notion, but it has a learning curve.
Prospects
This is early days for strong AI resources like ChatGPT and Notion’s AI Assist. There are many questions about their abilities. Their costs, too, which are much higher than Google searches, may limit availability of such services in the future. While there may be many practical benefits for this newest crop of AI, critical thinking is needed too. Students and faculty should avoid over-reliance on strong AI or overestimation of the capabilities of these new forms of mediating information. Faculty will also likely face the challenge of educating students about how and when to use this sort of resource, as they did with Wikipedia and Google. I also think we need to keep an eye on openness with such a powerful tool. (Here is an article that discusses just how open OpenAI really is and the need for openness in developing strong AI- https://theconversation.com/everyones-having-a-field-day-with-chatgpt-but-nobody-knows-how-it-actually-works-196378.)
I have been experimenting with input devices and methods for getting information into computers for a number of years. Many things have inspired this experimentation, but it mostly stems from the realization that much of my work as a teacher boils down to bookkeeping — recording grades and giving feedback on assignments. More and more of this work over the last 15 years has moved from hand-written grade books and criteria sheets to computer input.
Couple these changes with aging eyes and hands, and I am now looking for ways to make the process faster and easier. Hand pain for me has increased over time, and there are real health hazards, not just occasional pain, that all teachers and heavy computer users should be aware of. I have been aware of Carpel Tunnel Syndrome for awhile but researching the 2022 Collaborative Learning Days presentation confirmed that my main technique for managing pain and fatigue, varying input methods and pointing devices, can be a valuable ergonomic strategy.
This may not be for everybody, but I think varying methods is my main strategy for making computer input manageable and even enjoyable. What follows is a brief overview of techniques I use. This include both hardware and software that helps me input grades and feedback.
Pointing devices and keyboards
One method to manage computer input is to use different devices. I haven’t used many keyboards. I am pretty flexible in this respect, so I haven’t researched it much. There are many options for keyboards out there for better ergonomics, including split keyboards and various other physical arrangements. It just isn’t something that has caused me a lot of pain.
I did develop some index finger pain with standard mice, so I have tried a variety of pointing devices, including track pads and vertical mice. I once even tried doubling pointing devices, with a trackpad on the left and a mouse on the right side of my keyboard. It wasn’t for me. My current experiment is a vertical mouse, which is a mouse turned on its side, making the angle of the wrist more natural. The wrist angle is more like a handshake, instead of twisting the wrist downward. There are many cheap options for vertical mice, and they can be easily added to any computer.
Here are a few web resources on the ergonomics of keyboards and mice. The first is a guide from UC-Berkeley. The other two are buying guides with some good information on these devices.
Thanks to cloud storage, much of our work can be done on any device we like. I have been using an iPad Pro as a laptop replacement for awhile. It offers another way of inputting information, adding to the variety of input methods I use. With the iPad and the Apple Pencil, I have used it for hand-written annotation and converting small amounts of handwriting to text.
I recently discovered Microsoft Word dictation and it is surprisingly good. For some grade criteria sheets, I use annotation to write numeric scores and dictate longer comments, eliminating the mouse and keyboard entirely.
If you have a relatively recent iPhone, converting images to text is another way to input text. This can come in handy if you ever have text in an image that you want add to a document. While this has limited utility, I recently had a student email me a photo of a typed assignment, so it came in handy in this case.
https://lens.google – Android phones had this ability before the iPhone, called Google Lens
Boilerplate and clipboard management
Another kind of software approach to making input easier is to use boilerplate for repeated words and comments. These repeated words can be compiled in a document and copied and pasted, avoided repeated typing. There are also a number of applications that can store clipboard contents to accomplish this as well. Window 10 has a clipboard history function, which I just discovered. It stores 25 entries, and you activate it by selecting the Windows logo key and V.
This is a pretty thorough review of the latest version of Yoink and its capabilities on the iPad.
And now for a completely different kind of input device …
Finding ways to input information efficiently and ergonomically is important. But let me conclude with another kind of input that is important — drinking hot coffee. Grading requires some fuel and hot coffee is one of the best fuels I know, even though at my age I drink half-caf. Ember is nifty tech toy that replaces the old hot-plate type coffee warmers. It is a little quirky but I recommend this Bluetooth-enable heated coffee mug. Happy grading!
https://ember.com – Ember home page. These are available at Target and Amazon. It’s a little fussy and a bit pricy, but it is handy for “inputting” hot coffee or tea.
Kirkwood has recently adopted Pressbooks. This includes the ability to author H5P exercises, which can be used to enhance Pressbooks OER (Open Educational Resources) or can be embedded for a variety of stand-alone online activities. The H5P PB Kitchen site is a grant-funded project for the BCCampus (https://bccampus.ca/), which has been a leader in OER in Canada. Anyone who wants to learn about the use of H5P in Pressbooks could learn a lot by starting here. Alan Levine ran the project, whose grant has recently ended. The site still includes many good resources for learning more about H5P as a way to develop interactive OER.
Apple’s latest product announcements included a new camera feature for the 2021 iPad Pro, which could have value for faculty who record lectures. The front camera array has an additional wide-angle camera. The new Center Stage feature allows the camera to pan and zoom to keep speakers centered in the screen as they move around. (Update: I have seen conflicting reports about the placement of the front camera and whether it remains on the short side like a phone or has moved to the long side like a laptop. The PC World article mentions the change to the long side, but most others say the location is unchanged.)
According to bloggers at MacRumors, this feature will not be exclusive to Apple services like FaceTime, but should be available to third-parties like Zoom as well.
This article describes several startups trying to re-envision remote learning in response to the reliance on Zoom. Some services are Zoom replacements and Zoom add-ons. Others try to re-conceptualize the function that Zoom serves. Mid-way through the article, while talking to the founder of Top Hat, the author also discusses what the edtech industry thinks remote learning in higher education will look like after the pandemic.
I use Microsoft Office Ink technology for grade criteria sheets that mix typed text and written comments. This is a good use for the Apple Pencil, an iPad, and Microsoft Word. Using a rubric in Talon is often a better way to develop criteria and student feedback, but sometimes the criteria or the existing grade form doesn’t easily translate into a Talon rubric.
Requirements This method is just a form of annotation using Word, so you can do this with any tablet, stylus, and Microsoft Word. Depending on your tablet size, you may need to buy a subscription to Office 365 to use Word. I use a 12.9 inch iPad Pro with a keyboard and this hardware is ideal for this process.
Description of the process I start with an existing Word document. The only real modification needed for this speech evaluation form was to create text boxes where longer, hand-written comments usually go. You can see the box on the right.
Once I have a document to use as a template for student work, I save a copy for the individual students’ work. iPads tend to be set to autosave, which can be convenient, but not useful when you want to reuse a document. (I have not had good success using Word template files, with the extension .dotx, on iPads, so I use regular .docx document files. This may be a quirk with the file system on the iPad.)
Next, I fill out the form. I use the Apple pencil for short bits of information, such as numbers for scores. The text box is for more detailed feedback, which I type.
Once I have completed the grading form, I share it with a student by printing it or through grade feedback in Talon. Using a big tablet like the 12.9 inch iPad Pro makes uploading to Talon easy with split screen multitasking. This creates one window for Word and one window for a web browser.
In Talon, rather than type comments in the feedback box, I select “Insert Stuff” in the HTML Editor above the text input box. The icon for “Insert Stuff” is a little triangle in a black box, similar to the YouTube logo. The screenshot below shows the result of opening Talon in a browser window in split screen after clicking the “Insert Stuff” icon and choosing to upload a file.
Once the file is uploaded, you can label the link that students will see in the feedback box. The students will then be able to click the link, download the Word document, and view the feedback.
This is a good technique if you have the required hardware and software and you have a unique criteria sheet that is not easily reproduced in a Talon rubric. I like using the iPad as a change of pace anyway, so this is a good technique for me to give detailed feedback.
Here is a list of links if you are new to working from home or if, like me, you are new to teaching online. The first is an overview of online teaching considerations. The second is a link to a PDF-formatted email newsletter on my Dropbox account from the Chronicle on emergency online teaching. The third is an overview of using team-based learning online. The last is a freebie ebook offer from the Take Control book series on working from home.
You must be logged in to post a comment.