EconBot: a chatbot to support the online learning of Economics
Perhaps one of the most complicated tasks for the instructor in distance university teaching is related to motivation and student support. For students the learning process usually takes place in an environment of isolation or solitude, mitigated by the availability of different communication tools within a standard LMS, and more recently by the diffusion of the use of social networking and instant messaging applications.
From the end of 2016, and throughout 2017, it seemed that bots came to facilitate this support function. The opportunities that bots could offer for education in the immediate future were exposed very clearly here, here, and especially with the case disseminated by Ashok Goel from the University of Georgia.
Meanwhile, a large part of the bots that companies dedicated to electronic commerce began to use did so through the most widespread social network -Facebook, specifically through its Messenger application, developing conversational bots or chatbots.
It is in this context that the possibility of developing a chatbot for the subject “Economics: Microeconomic Fundamentals”, taught in the third year of the Degree in Social Work, was proposed in May 2017 at the Department of Applied Economics of the UNED.
Motivation to create a chatbot in the University
What was the main reason for adopting this decision? In spite of the homogenous character of the organization of learning times in the Spanish university system, patterns and forms typical of certain Universities still survive. This is the case of the University where I develop my activity, where students have an exam option that takes place at the end of the summer (in the first week of September), while ordinary exams are held at the end of May and at the beginning of June. The student who does not pass the subject in that ordinary time has an additional time of almost three months to prepare his exam again. And, during those three months, the professors are not obliged to attend the forums of the subject, nor to attend the rest of the means of support for the student: at the end of the semester, my student care activity ends. In summary: during the development of the semester the student has a wide range of regular means of communication or support for their study, something that disappears during the summer time, so that the student does not have the same level of support or communication.
So the objectives set for the design of this new tool were the following:
- Have, experimentally, a new way of support for the student, precisely at that time when the other tools are no longer available.
- Evaluate the real possibilities, beyond the hype of the moment, of this type of tools to improve the teaching / learning processes of Economics in non-economic social degrees.
- Have a good time experimenting with the personality and style of a new “partner” in the teaching of Economics.
How do we create EconBot?
Having no experience in programming to create a chatbot from scratch, and for reasons of economy in the use of our time, the first task of the project consisted of selecting the software to use for its design. A search was carried out on existing web services to design and maintain chatbots, about their advantages and inconvenience, to finally opt for the use of Chatfuel.com.
Chatfuel.com is an intuitive web platform, which has improved its functionality progressively, and with very attractive design possibilities in its free version. One of the few “cons” that can be pointed out is that, at the moment, it does not allow downloading all the arguments that have been introduced in the “Artificial Intelligence” section, so that if at any moment the chatbot is to be replicated in another platform , or with specific code, all those arguments should be rewritten. All in all, Chatfuel allows all those arguments to be duplicated in a new bot, which is very useful when you have built a large library of “small talk” that can be used in a new chatbot.
The choice of Messenger as the application for which to build a chatbot was simple. Its great virtuality is that it is part of the Facebook ecosystem, widely spread among students, and in its widespread use through mobile devices, which gives it a clear superiority over other communication tools contained in LMS. In addition, in the course we were already using Facebook as a support social network in the distribution of information in open so it seemed logical to place within that ecosystem the choice of the chatbot.
So building our chatbot, named EconBot, began (obviously expressed only in Spanish) with the intention of:
- Send periodically during August support messages to the students, with indications of concepts, ideas or economic relations especially important (actually, information similar to that contained in the guide for the course, but our experience shows us that students do not usually read …). Links were also sent to available support materials, such as mind maps.
- Periodically send memes, images, or gifs related to important concepts and ideas. For example, a gif that shows the difference between displacement of demand, and movement along the demand.
- Send every two days an exercise or quiz question, so that the student will answer them throughout the day. Late in the afternoon, EconBot would inform each student of their success or error, with a brief explanation of the related terms.
The launch of EconBot
Once the availability of EconBot was communicated by email to the students, 30% of those who could be examined in September clicked the “I like” option on the Facebook page created to make the chatbot known. During the month of July, 21 EconBot messages were scheduled to be sent during August to all the students who accepted to receive them in the option offered to them when they began their first conversation with EconBot (a subscription menu, to which they could access at any time to unsubscribe).
For each exercise or quiz question EconBot stored the answers of each student, which allowed to segment subsequent sendings with reinforcement messages specifically aimed at students who had failed certain problems or quiz questions. This is the main advantage of a chatbot with the chosen design: the enormous possibilities it offers to personalize the support to the student depending on what their answers to the different questions or exercises raised from the chatbot have been.
A base of arguments and expressions was also built, to provide EconBot with basics means for a “small talk”, and it was enriched during the month of August when checking that many of the users were trying to hold conversations with him, and that EconBot did not always have the right answers. The themes and the tone of the conversations were very diverse. Sometimes the students asked him to explain a concept or a specific economic relationship, for example, the opportunity cost. In those cases, the default response available to the chatbot included recommending them to use the webapp of the subject that includes a Glossary with the basic terms for each of the lessons of the subject. However, we have found that students can find it more attractive, and possibly more effective for their learning, if the question about a specific concept or relationship is answered in the form of a conversation, in which the chatbot address the student by her name.
The main advantage of a chatbot is the enormous possibilities it offers to personalize the support to the student.
On other occasions students asked for the solution to specific exercises included in the textbook of the subject, something for which the EconBot was not, and still is not, prepared.
EconBot is equipped with a main menu from which the student is offered access to other existing learning resources, such as the webapp, the course in iTunes U, or the YoutTube channel where the explanatory videos of the subject are collected. This was to provide them with mobile access to resources whose existence the student already had to know.
Unfortunately, given the anonymous nature of the followers of EconBot, and the technical and legal impossibility of relating these data with those of our real students, we can not have any evidence on the effectiveness of this type of tool in the improvement of the learning results of our students.
Some conclusions
From our experience we believe that the role of what we call “heroic instructor” should be valued very positively: one who dedicates his time and efforts to designing this type of technically complex tools, thus showing initiative capacity in the adoption of innovative resources for teaching and learning. Although for these teachers this type of initiative tends to have a high opportunity cost in terms of other teaching activities, or of their own research agenda. Therefore, the ideal scenario must be one in which universities have specialized units in the design and development of this type of tools, which in addition to saving scarce and precious time for instructors, can generate final products of better quality technique.
However, it is not risky to say that chatbots, whatever the concrete form in which they evolve, will be more than a passing fad in the education environment, and that they will probably play auxiliary tasks in educational processes, providing support to student at a distance environment, at any time, and on any aspect related to their learning activity.
Once the month of August was over, we decided to keep the project in “hibernation”, waiting to finish a new regular teaching period, to resume it again now with the intention of reusing it as a means of supporting the student during next summer, now enriched with the experience of the last year.
We intend to create conversational modules for the most relevant concepts, ideas and economic relationships, using the principles of design of multimedia learning materials that have proven to be effective, and with which we can use this chatbot in the future during the ordinary time of academic activity.
It is also our intention to continue with the development of a comprehensive and enriched module of “small talk”, that allows to simulate in a credible way a conversational relationship between EconBot and our students.
