Conversational Agents

Doering et al. (2008) found that while conversational agents engaged learners outside of the work that they were doing, they were not very helpful in developing their projects (p. 267). Similarly, Veletsianos, Scharber, and Doering (2008) state “educational technology researchers appear to be overly enthusiastic regarding the possibilities afforded by pedagogical agents, even though it appears that there is no compelling experimental evidence for their learning benefits” (p. 292, Choi and Clark, 2006). In addition, both studies refer to the sexual content that users injected into their conversations with the agents. De Angeli and Brahnam (2001) found that “approximately 11% of user–agent conversations were concerned with hard-core sex”. I find this troubling, but not surprising. Many middle and high school students are going through puberty and beginning to think about sex. Also, technology makes it easier for many to say what you are really thinking. Would most students approach a teacher and ask them about sex? Probably not. Through text messaging, online friend services such as Facebook, and online chatting many adolescents and young adults express themselves in ways that I believe they wouldn’t do when in face-to-face situations. I think that there are often fewer emotions associated with the written word. I would argue that if you asked a student to ask a vulgar question, the chances of them doing it are higher than if they had to speak it.

I think that as they are currently, conversational agents have little educational value, however this could be changed with improved agents. Doering et al. (2008) believe that “if learners are to utilize CAs with success, the CAs need to be intelligent enough to accurately comprehend the questions the learners are posing or the software application needs to offer an interface that guides the learners to ask their questions in an appropriate manner” (p. 267). Agents are not going to be valuable in learning situations if they do not answer the questions that the students want answered. However, Doering et al. (2008) found that agents had other benefits such as aiding in student reflection, helping students think about questions, discussing issues with students, and dialoguing (p. 268). With better, updated agents, they will be very useful in the classroom.

One idea for classroom use is to set them up for students to use while there is a substitute teacher. While working on a paper or project, the students could use agents to answer their questions, as well as the substitute. Similarly, they could be used after a unit, and before a review for students to ask any questions that they might have, as well as have the agent ask students questions, much like a quiz, but more fun and conversational. I do think that agents could be very useful for learning situations if they were really good agents, with a lot of knowledge. I think that the fact that they have personalities will make students want to converse with them, aiding in their interest in the task at hand. Another idea is for students to set up agents, much like we did for our blog, as part of a presentation. This could be especially useful for shy students who often stress over speaking in front of groups. It could be given as an alternative to speaking themselves. Though this does not teach speaking skills, when students are severely stressed out, they do not learn either. I was this student in middle and high school; the student who could not speak in front of a large group. I eventually grew out of it. However, on my way to growing out of it, I avoided every class that involved speaking in front of others. Had something like these agents been an alternative, maybe I would have considered those classes and learned a thing or two about writing speeches.

Overall, I think that agents are a cool, but also totally creepy thing. In the future, with more development, they could become very useful in the classroom. Also, they were fun to set up, though time consuming.