Take My Advice: If You Have Trouble With Tech

Sometimes technology betrays us! It’s happened to us all, usually at a point when we are actively trying to seem competent. If you ALWAYS have problems, however, there might be something on your end you could change. There are some common problems I’ve identified from my time supporting faculty and graduate students with technical issues; these issues not only also affect students (and thus could be usefully passed on to them for their own tech needs) but also characterize a decent portion of my own issues when I stop to reflect. Some of the biggest below:

You’re going too fast. Technology can often allow us to go faster! But going too fast with a technology can not only lead you to make mistakes (guilty), it is also sometimes simply too fast for the tool to keep up. For example, Canvas often needs a refresh to show an updated grade when grading quizzes– it simply won’t update its view for you immediately upon the updating of a quiz grade. And, it’s of no use to start reviewing New Analytics in Canvas immediately; it takes time for it to analyze student activity and grade data and present it to you. It can help if you think of technology not as an instant communicator, but as a collaborator who might take a little time to get back to you. 

You’re using the wrong browser/old hardware. Imagine the software you’re trying to use is a gold-plated, top of the line refrigerator. It’s very nice, it functions perfectly, and it has a lot of features that you’re looking forward to using. Now imagine you give that refrigerator to a Pekingese and tell it to carry it up the stairs, install it, and get it functioning.

A Pekingese who would try really hard, though. Image by No-longer-here via Pixabay.

Not only is that Pekingese not going to be able to lift that refrigerator, it’s also not going to have any idea of what you’re talking about. Using incompatible hardware or software to get your tool working is similar– generally, it’s either not strong enough to carry the task, it doesn’t have the capacity to understand what it would need to understand to carry out the task, or both. 

You’re using a nonstandard tool for the job. Imagine that refrigerator again. It has a door, right? It plugs in. You put food in it. All these qualities are very similar to a microwave, but if you try to use the refrigerator as a microwave, you’re going to be disappointed. Similarly, looking at what a tool was designed for and evaluating if that is in line with your ambitions for the tool can be helpful. That’s not to say you can’t use a tool at sideways purposes– I would argue half of my discussions of Twine are sideways to its intent. (Incidentally, the medium of Interactive Fiction also has a lengthy history of using tech in ways contrary to its intention for creative purposes, so I feel I’m part of a rich tradition.) However, you’ll experience less frustration if you do so intentionally, rather than realizing that you’re trying to give feedback with a software that’s really only meant for file sharing, or facilitate group collaboration with a tool that only allows one or two users at a time. 

I’m in an interesting predicament at the moment where I’m trying to use AI tools to complete some data analysis. I am reading about good prompting strategies and trying to use them. I’m trying to build a foundation and then build on top of it. I’m trying multiple different strategies to see if I can come to some different results, even if only to give me some insight into how to better get the results I want in future. Yet, at the moment, it’s been more time checking and evaluating and figuring out where it’s getting its ideas than actual time saved in the analysis from the way I normally do it. 

I don’t think I’ve been going too fast, although I admit sometimes I see the demos of how quickly AI can pop together information and read reports of how it’s dramatically cutting time spent on certain tasks and I’m like “shouldn’t it just be able to do this for me already??” My hardware and software don’t seem to be the limiting issue– it’s not that it’s too slow, it’s that the conclusions are incorrect or partially correct or inventing some information that is not quite there. And the tool by all reports seems to be the tool for the job (see above demos), although perhaps a lesson I may end up learning is that some parts of the job are beyond it. I think the issue is actually a number 4, which is that the tool is more complicated to use than I am treating it as, and I don’t know enough yet about using it effectively. Straightforward and simple UX is powerful, but it can come at the expense of tailored, expert usages– think about what can be achieved by a pro using HTML to create a website versus a novice using Blogger or WordPress, or by a musician versus the prerecorded tracks on a child’s toy keyboard. The problem I’m trying to solve, then, may simply be one that is more “improvisational jazz” than “Old MacDonald.”

 

The strategy that's been most helpful to me so far is one that can be difficult for a lot of us to use: ask someone else to make an attempt at it, and learn from what they do. When in doubt, have someone else try! Either they have success, which you can learn from, or they don't, in which case you can at least feel a bit better that you are not alone.

Any of these tech issues plaguing you, or others? Let me know in the comments!

Plug and Play Miniseries Lesson Spotting with AI: Designing Prompts (but not like that)

Photo by Işıl via Pexels.

In my last feature in this series, I talked about using GPTs to help in the process and clarification of your writing. This entry is a bit of a cheat in this series, since it doesn't actually involve prompting a GPT at all. 

Good UX always inspires me to think about how to use some of these strategies more broadly. I remember being mildly affronted during the CITL reading group I attended when I first became interested in better teaching-- did I spend that whole several weeks affronted? Apparently! Clearly it was challenging me in some useful ways-- because one of the resources we read discussed using advertising principles to make learning "stick."  Advertising? That soulless capitalist enterprise? Could help me teach the intellectually rigorous discipline of history? Pish-posh! 

Of course, the only thing we do if we eschew these strategies is ensure that what they're learning in our class presently doesn't seem quite as memorable as literally any reasonably well-crafted local commercial they saw roughly a decade ago, which is what I eventually realized as I considered the concepts that were "sticky" in my brain and why.

With that in mind, you might reflect on ways that new tools introduce themselves to you, and see if that sparks an idea about how you might in turn introduce a concept that is very familiar to you in a way that seems approachable to a new learner. I'll use the original starting page for ChatGPT as an inspiration to explain something, such as instructions or expectations for an assignment. One thing that is inherently challenging is to explain something you know very well to someone else. It can be more challenging than explaining something you only know kind of well., as you have to identify the most significant information out of all the information you know and explain it clearly.

Let's take a look at how ChatGPT introduced itself when I encountered it (this has since changed as public familiarity with the tool has increased):

Original ChatGPT intro screen. Image via Datamation.

This screen introduces the tool by offering a breakdown of examples of things you could ask it ("Explain quantum computing in simple terms"), capabilities the tool has ("Remembers what user said earlier in the conversation"), and limitations the tool has ("May occasionally generate incorrect information.")

What if you tried a Examples, Capabilities, Limitations breakdown for assignment instructions?  I've seen many, many pagelong or multipage prompts for a paper that's only 3-5 pages long. Rather than paragraphs of context and/or admonitions based on past experiences ("12 point font and 1.5 inch margins this time-- I'm talking to you, Bradley"), what might it look like to organize a prompt arount the following structure: 

1. Examples (of excellent work or strong approaches to the work),
2. Needed Capabilities (of the deliverable students will produce), 
3. Limitations (or parameters they need to work within to produce this work)

Obviously a prompt doesn't have to stay in this format-- I'm not suggesting that the original ChatGPT welcome page cracked some kind of fundamental educational code. But instructions that you find clear or helpful in introducing you to a new topic may be useful in turn to incorporate into your own teaching strategies.