My Experience with Introducing AI in the Classroom

By: Raed Jarrah

I like to think of Artificial Intelligence as just another tool in the vast arsenal of applications that construction management students will be using in their line of work. I have started to introduce AI in two of my classes this semester: “Developments in Construction Technology” (CNST616) a graduate course, and “Construction Law” (CNST406W), a writing-intensive undergraduate course. In both courses, I encourage the use of AI to help draft assignments.

When updating “Developments in Construction Technology” for this semester, I intended to cover AI as one of the topics. I wanted to share with the students how this technology is helping managers draft emails, write reports, summarize minutes, and much more. I wanted to explain to the students that this is a really cool new tool that people in our industry are just starting to utilize. And that's when I realized it would be very hypocritical to show students how AI is being used in our industry but bar them from using it to write their assignments. 

For both those courses, when we were discussing assignments on the first day of class, I asked: “Are you allowed to use AI like ChatGPT for your assignments?” The response from students was an emphatic, “Of course not!” I then asked, “Are you allowed to use ChatGPT at work, like to draft emails?” and the response was a more timid “yes, maybe?” That is when I explained that those tools are allowed in my class just like spell-check, Google scholar, and Grammarly. We then briefly discussed how AI can be a good tool for first drafts, but students should always review the facts, as they’ll be responsible for what they submit. I mentioned, to emphasize this, that two New York lawyers were fined for submitting an AI-drafted legal brief with fabricated cases.

We’ve had some interesting experiences with AI so far. In the graduate course, the discussion topic about AI in construction was scheduled to be the same week when term paper proposals were due. The week before our discussion, I asked students, “Who has been using AI…,” and a few students raised their hands, but when I continued, “…in drafting their proposals?” all hands fell. I encouraged them to explore the upcoming discussion material as they worked on their proposals. The week after, when we met for the AI discussion, I started again with the question, “Who has been using AI to help draft their proposals?” and almost everybody raised their hands. 

The following week, I asked the students for their impressions on the day’s topic, AI. One international student said she was blown away by how ChatGPT could eloquently rephrase her thoughts from Spanish to English, helping her organize her paper much more efficiently than brainstorming with a roommate and using a Spanish/English dictionary. Another student shared that he tried out ChatGPT to draft an email to a subcontractor gently declining their expensive offer. He too was amazed at the output, saying, “It came up with many eye-opening points I had not thought about” (like encouraging them to apply in future bids). He was also impressed that it made it more concise on command. This student clearly demonstrated using AI in a professional application. Moreover, both students acknowledged that the initial AI-generated drafts needed further refinement and review.

For the first writing assignment in the Construction Law class, I got the impression that students were not using AI. However, one student submitted an assignment that had all the telltale ChatGPT signs: 

1. Completely misunderstood the legal case.

2. Confidently presented irrelevant information.

3. Fabricated references and citations.

4. Ended with a superfluous “conclusion.”     

Some citations weren’t in the references list, and some of the references were not cited (I often see the former but almost never the latter). One of the fabricated citations was being used to support everything: irrelevant legal concepts, mentioning a plaintiff’s name, and even the student’s own personal opinions. 

While there is controversy around AI detection software, I plugged in this student’s essay in three AI detection platforms out of curiosity. Two platforms returned a very low probability that it was written by AI while the third indicated a " very likely” result. That’s probably because there were some blatant spelling errors in the document. I believe the student put in an honest effort in revising the draft; the first part of the essay correctly presented ethical concepts we had discussed in class.

I started my feedback to the student with appreciation for trying new tools, but also pointed out that these tools have flaws. I graded the assignment for the student as it was, using the usual format of showing where points were deducted and the reasons for the deductions. As with most of my assignments, I informed the student that the work was a good start but was not quite where it needed to be, so they have the option to revise and resubmit it. Within five hours, the student had adjusted those points, provided proper references with citations, and submitted a paper that earned them a full grade.  

Introducing AI to the classroom has been a new experience for me and my students. I like to think this is the next new tech revolution, and knowing how to utilize it will become an important skill. Writing correspondence has become much faster as spell-checking evolved into auto-correct and now into auto-complete. Meetings (ugh) have evolved from flying people to attend in-person, to huddling over a lagging telephone conference, to video chatting in comfortable pants. I am excited to see where this will go. 

In conclusion, I for one welcome our new AI overlords (Smith et al., 2017)