Using Chatbots for eLearning

Interaction, engagement, and personalisation are often cited as elements of the ideal electronic learning experience. And there are various methods by which these ideals are incorporated into eLearning courses, such as gamification, collaboration, and multimedia environments.

There’s an evolving technology which has the potential to add to this mix – one which combines information dissemination with entertainment and a personal touch: the use of chatbots.

What is a Chatbot?

A chatbot or conversational agent is a piece of software which simulates human-like conversation through artificial intelligence. Chatbots are often deployed as virtual assistants, and may have a visual representation (often in the form of a cartoon character or photo-realistic animation), or else be a disembodied voice or text stream. Microsoft’s Cortana, and the smartphone virtual assistant Siri are common examples.

Communicating with Characters

Usually, chatbots are pre-programmed with a database of words, phrases and questions to which they’ll give a pre-defined response. Some may also chip in at random intervals with jokes, quotations, or helpful hints on tasks currently being performed (such as driving instructions, based on your GPS location).

communication bubble

They may use text-to-speech technology to deliver their responses as simulated human voices, or display a text box or speech balloon that contains the words being “said”.

In turn, the user may communicate with a chatbot in several ways. Text may be entered into a form field, or the user may select an option from a multiple choice set of questions or statements displayed on-screen. And if a system is configured for speech or voice recognition, the user may communicate with the chatbot vocally.

Clearly, this input and response process has engagement and interaction on the part of the user built in, and chatbots have already found applications in several areas.

The Sofia Experiment

In 2003, the Mathematics department at Harvard University in the USA conducted an experiment involving an intelligent chat agent called Sofia. The program had an encyclopaedic database of mathematical definitions, a background in general knowledge, and could give the solutions to simple mathematical problems.

Students were given the opportunity to expand their mathematical skills with the aid of Sofia, which recorded all conversations for later analysis.

The results of the experiment showed benefits on all sides. Students surveyed found that by training the chatbot and watching how Sofia “learned”, they gained a better understanding of mathematical processes and teaching methods. And Sofia’s archive of recorded sessions built up an open resource for different branches of mathematics, while shedding light on how the students themselves learned, what their problem areas were, and what questions they were asking.

Use in Corporate Settings

corporate buildings

There have been notable applications of chatbot technology in the corporate sector, too.

Flat-pack furniture phenomenon IKEA uses a chatbot named Anna to interact with visitors to the company’s website. Customers can navigate the Web resource by entering specific questions about what they need, and following the directions that Anna gives, in reply.

The Real-time Internet Technical Assistant or RITA is used to assist customers in effecting wire transfers and other transactions at the ABN AMRO Bank.

And a chatbot named Abby is deployed by the company GetAbby as the central part of a Customer Relationship Management (CRM) system that uses voice recognition to track and log information from telephone calls, including customer details and the conversations themselves.

Deployment in Security Training

In a study conducted by researchers at Sweden’s Stockholm University and Royal Institute of Technology (Stockholm), 26 workers from a large international company and 80 of its security specialists were given a corporate training course on internal security. Those in the group that worked with a chatbot named Sally recorded a more engaging and satisfactory eLearning experience than those who did not.

Integrating with Social Media

social media symbols

Some researchers in the field of chatbots have suggested a logical progression to their use on Web resources and social media platforms. As an instance cited, Google’s ongoing development of its text mining and Natural Language Processing technologies could extend to putting a face and personality on the search engine’s output, increasing its accessibility and appeal to users. And the large communities and app capabilities of platforms like Facebook could be used as training and testing grounds for chatbot programs.

Chatbots in the Wild

Third party providers are getting in on the act, with pre-packaged or custom-tailored chatbots available for hire or purchase.

Organisations like Guile3D have an extensive roster of their own photo-realistic avatars, which can be bought and downloaded as virtual assistants, entertainers, and digital spokespersons. You can also provide a list of specifications to modify existing characters, or create your own.

And the Chatbots.org website is an online portal and resource for chatbot technology and related applications. So if you’re in the market for a desktop William Shakespeare (who tutors in English Literature, naturally), there’s a place for you to go.

Some Design Tips

• If you intend to use a chatbot as part of your eLearning course, you’ll need to consider the nature and demographics of your intended audience. This will influence the syntax and tone of responses and prompts that your chatbot should give.
• If you’re using a visual avatar or animated character, the chatbot’s appearance also needs to be consistent with the audience that it’s addressing.
• Use existing course notes, interview transcripts, and clips from related media (press reports, professional literature etc.) to boost the chatbots database of knowledge and vocabulary.
• Lists of Frequently Asked Questions (FAQs) may also be used to generate a chatbot’s list of pre-packaged queries and responses.

Tags: , ,