Keith Weber

I have taught management information systems at the RIT Rochester and Weihai campuses since 2018. Fun fact: I consider teaching to be my sixth career, after several others, and being empty-nested by my adult children.

1. How do you use AI in your teaching, and what are your favorite resources?
 
I have found GPT4 and Bard to be great resources when brainstorming lesson plans and assignments. For example, I often prompt them with what I am teaching and ask for the salient points I should cover. Most responses are obvious, but I might have missed some points. I ask them for questions that someone may ask in discussions and breakout room activities about given topics. I will have them ask me questions that students might have about these topics and have the chatbots challenge me to answer. I will prompt them to give me business scenarios that fit specific criteria to provide varied and interesting fictitious teaching cases to analyze and solve. I use Bard when looking for information dependent upon up-to-date sources or how to accomplish something using specific cloud or desktop apps. When I need the system to provide me with information that is less about regurgitating the facts or generally accepted conclusions liberally sprinkled around the internet and more ‘considered,’ then I find GPT4 to provide better output. For example, it provides much better business model canvases (BMCs) as starting points for IS design activities (see below). I have found Claude.ai and Perplexity to occasionally fill shortcomings from other textual generative AI (“genAI”) tools. Bing has left me generally unimpressed. I have not done any substantial experimentation with image-based genAI tools. It should go without saying that I do not swallow anything these tools produce hook-line-and-sinker, of course, but – frankly – as brainstorming and grunt-work automation tools, I am sold on their value. See my discussion below about what I tell students for thoughts on the burning question of output quality.

2. Can you share or describe an example or two of an AI-related assignment?
 
Because I teach on the technology side of the business college, I have limited time to coach students on subjects such as business models and strategies. However, the output of those analyses are important inputs to information systems design. I will have students prompt for business model canvases for specific businesses or types of businesses to generate some of these details. This gives me an opportunity to discuss the roles and importance of these inputs without digressing too far into subjects covered by other business college courses.

3. What do you tell students about using AI?

In my opinion, the best antidote to the risks of genAI is for users (students, researchers, businesspeople, and practitioners) to understand what it is and is not. At its most basic, text-based genAI automates something we have all done repeatedly for decades. When we wanted to uncover facts or understand something, we would conduct an internet search and then click through to (often) multiple pages looking for answers. We would often subconsciously judge the sources, collect the facts or evidence, and, finally, come to conclusions. Text-based genAI does something very similar for us, but by collecting its information from sources that we do not lay our eyes on, it denies us the opportunity to accept or reject individual sources. I explain to students that this is not all that different from what happens when we look up something on Wikipedia. Most of the time, it is reasonably accurate. It can arguably be used to inform and refine our understanding of non- or semi-technical subject matter. But, as many of us know, even when it includes citations, those citations are often to less-than-stellar sources. Wikipedia is good for what it is good for, and a savvy consumer of information uses it for its strengths while distrusting it for purposes that have a critical need for accuracy and authority. GenAI can save us enormous amounts of time producing the type of preliminary matter that goes into many non-scientific work projects. By providing detailed responses that can span pages, its output resolves one of the prime challenges of human intelligence: the limit on what we can hold in our short-term memory at any moment. By putting it on a page or two, we can consider a bigger picture than we can assemble in our short-term memory. We could (and used to), of course, accomplish this ourselves by doing the preliminary data gathering and organization the old-fashioned way, but now we can, within seconds, have all of that done and laid out neatly in the text on a page so that we can move on to apply our human intelligence to it. GenAI can perform much of the preliminary analysis and evaluation, especially if we dialog with it and challenge its answers. This is just the beginning of a much longer conversation I have with students, but my point is that we cannot supervise what we do not understand. Demystifying genAI, removing it from the realm of computational magic, and recognizing it as automation of many operations we have been doing all along is step one.

4. What challenges, if any, have you had with AI in your courses?
 
I can no longer trust that anything actually comes from my students unless I see them produce the work in front of my eyes. This has caused me to reevaluate the types of assignments I give my students and to reemphasize those that involve the public production of student thought and innovation.

5. How do you think AI has or will impact your domain?

I’m glad you asked! I am running an RIT TLC Teaching Circle this fall (2023) to explore this question for various disciplines across the campus. Please email me if you would like to join us as we dust off our crystal balls and figure out how genAI (both text-based and image-based) might affect the software tools and job market for our graduates. kweber@saunders.rit.edu