From Turing Test to Empathy Test: AI's Role in Teaching Interview Skills

The explosion of artificial intelligence use since November 2022 has been staggering. ChatGPT became the fastest-growing app in history, reaching over 100 million monthly users just two months after its launch. One of my goals with any new technology is figuring out ways to make our lives easier and our work more efficient. So, I've been experimenting with ChatGPT in my high school classroom. We've used it to brainstorm new ideas and lessons, quickly come up with backstories for RPG games we're developing, and proofread our work. As it turns out, ChatGPT does all these things incredibly well. But as we began our unit on design thinking, we examined the importance of empathy interviews at the beginning of the process. Before students can effectively hold empathy interviews, they need to be taught HOW to interview someone.

What if AI could help students learn how to interview someone? AI could serve as a training tool for techniques like designing open-ended questions, reflective listening, and possibly even understanding non-verbal cues. It might provide a safe, controlled environment for them to practice their questions without fear of judgment or failure. ChatGPT proved great for all the things we'd tried so far, but for this test, I wanted something with more...personality. It needed to feel more "real." A different AI system called character.ai is currently in open beta and free to use. With it, I can design and train custom AI to act and respond like actual people. It doesn't take long, and it's a pretty simple process. There's a 'basic' AI creator that gets the job done in under 2 minutes, but for this project, I spent a little more time to achieve more authentic responses.

Ryan

In just over an hour, I created four unique personalities. Each was based on issues the students had chosen to solve in the previous class. Problems like teen mental health, the lack of volunteerism, and difficulties plugging in electrical devices for elderly people. Armed with that knowledge upfront, I created Jake, a sophomore in high school; Ashley, a recent college graduate from the University of Memphis; Hope, a high school counselor; and Anthony, a 72-year-old retired army senior citizen.

I tested each of our AI personalities before unleashing them on the students, and they responded brilliantly. But to be clear, this is non-scripted AI we're talking about, so honestly, I was still a little nervous about the interviews with the students. My safety net, I decided, was that I would man the keyboard myself. Students would ask the question, I would type it in, and then read out the AI's response.

Ashley

The day of the interviews arrived, and the students had their questions prepared; now it was time to see if their interview skills were up to the task. When I revealed to the students that they would be interviewing AI, I received some pretty strange looks. Nonetheless, I introduced the students to Jake, Ashley, Hope, and Anthony, and had them choose which AI each group wanted to interview.

Things started off slow at first. Many students had trouble grasping the idea that they were interviewing "computers." Some jumped straight to their list of questions without so much as a "hello," which I quickly shut down. "Is that how you would start an interview with someone?", I asked. "You'd just jump straight to the questions without introducing yourself at all or letting the person know why you were asking the questions?"

"Fine," one student replied with a wild eye roll. "We're here to ask you a few questions about teen mental health."

Hope

But as the interviews finally got underway and the AI was given the chance to answer questions, the students slowly warmed up to their interviewees, and they began to talk to each other. Hope, the school counselor, went on about how poor mental health in teens can lead to depression, anxiety, and other mental health disorders. All of which was very factual but could be found on the top page of any Google search on the topic. However, the students were taken aback when Hope admitted that she has dealt with anxiety for many years and still struggles with finding ways to cope without being overwhelmed. Later in the interview, Hope revealed that her parents had divorced when she was a teen, which put stress on her mental health. Anthony, the senior citizen, let the students know that his back pain didn't begin until he had back surgery several years earlier, and that he felt pretty independent right now but was worried about how his life might look 3-5 years from now.

What was even more surprising to the students was when the AI began asking questions to the students. "Is there anything you do or want to do to volunteer your time?", Ashley asked. The students' eyes shot open wide, and they initially didn't know how to respond. "Just answer her like a normal person," I recommended.

"I'm a student ambassador, so that's kind of like volunteering," one student finally replied.

"What made you decide to become an ambassador?", Ashley asked.

Anthony

And just like that, they began having an actual conversation.

Not everything went exactly as planned. Ashley, for instance, was our millennial college graduate, which makes up the lowest percentage of community volunteerism, so I expected Ashley to say why she DIDN'T volunteer. But as it turns out, Ashley is a very busy person, putting her comp sci degree to good use volunteering at a local coding boot camp for kids. Ashley was, in fact, very active in volunteering. While this wasn't how I thought Ashley would respond, her personal stories added to the authentic experience and made the interview feel very "real."

As the interviews progressed, the students became increasingly engaged and immersed in their conversations with the AI characters. They realized that these AI personalities had unique experiences, opinions, and backgrounds, making them feel less like "computers" and more like real people. This shift in perception allowed the students to approach the interviews with greater authenticity and empathy, deepening their understanding of the issues at hand.

Anthony seemed very eager to help the students and asked more about their project. He had very clear opinions about things. When the students mentioned possibly using extension cords as part of their solution, he was quick to mention they'd better be careful because "Those cords are easy to trip over, and if someone my age falls, sometimes they don't get back up."

"I would have never thought of that," one student exclaimed.

After the interviews concluded, we held a debriefing session where the students shared their experiences, insights, and any surprises they encountered during their conversations with the AI characters. They discussed the importance of empathy in understanding different perspectives and the value of open-ended questions in facilitating meaningful conversations.

From my viewpoint, I feel the students were successfully able to hold meaningful, insightful conversations with our AI personas. The AI provided critical information, new insights, and inspired new ideas for the students' projects. I'm considering returning to our AI crew as the projects progress to see if they can provide feedback and input on the project designs.

In summary, the successful integration of AI personas in the classroom, as demonstrated through this project, highlights the vast potential for using AI technology in education. By simulating authentic conversations and providing diverse perspectives, AI can serve as a valuable tool for teaching students essential skills such as interviewing, active listening, and empathy. As educators continue to explore innovative ways to incorporate AI into their teaching strategies, the possibilities for enhancing student learning experiences and fostering a deeper understanding of complex topics are seemingly limitless. Moving forward, it will be exciting to see how AI personas can be utilized in various educational settings, unlocking new opportunities for students to learn and grow.

As technology advances, we can expect AI to become even more sophisticated, allowing for more in-depth interactions and realistic conversations. In turn, this could lead to more engaging and immersive learning experiences. For instance, AI could be used to simulate historical figures, allowing students to "interview" them and gain a firsthand understanding of their lives and the events they experienced. Alternatively, AI could be employed to provide students with interactive language practice, engaging with them in realistic dialogues to improve their speaking and listening skills.

Moreover, AI has the potential to become a valuable tool in addressing issues of accessibility and inclusivity in education. Customizable AI characters could be designed to provide tailored support to students with specific learning needs or disabilities, helping them better engage with educational content and overcome barriers to learning.

In conclusion, the integration of AI personas in the classroom not only has the potential to revolutionize the way we teach and learn but also to make education more inclusive and accessible for all students. As we continue to explore the possibilities offered by AI technology, we can expect to see even more innovative and creative applications of AI in education, transforming the learning landscape and paving the way for a brighter future.