What are the best practices for designing conversational agents for online learning?

2 Choose the right type and mode

There are different types and modes of conversational agents that you can use for online learning, depending on your learning goals, content, and audience. For example, you can use a pedagogical agent, which is a virtual character that acts as a tutor, mentor, or coach; a social agent, which is a virtual character that acts as a peer, friend, or companion; or a domain agent, which is a virtual character that represents a specific domain, such as a historical figure, a scientist, or a fictional character. You can also choose between text-based or voice-based conversational agents, or a combination of both. Text-based conversational agents are more accessible, flexible, and scalable, but voice-based conversational agents can create more natural and engaging interactions.

3 Design the dialogue and interaction

The dialogue and interaction between the conversational agent and the learner are crucial for the learning experience. When designing them, you must consider the tone and personality of the conversational agent. What traits, values, and emotions do you want it to have? You can use humor, empathy, praise, or feedback to build a positive relationship with the learner. Additionally, think about the content and structure of the dialogue. Will it match the learning goals, content, and level of the learner? You can also use branching, adaptive, or generative dialogue techniques to create dynamic conversations. Lastly, consider the interaction design and feedback. What cues, prompts, and signals will the conversational agent use? How will it provide feedback and support to the learner? Multimodal input and output such as text, voice, images, animations, or gestures can help create rich conversations.

4 Test and evaluate

The last step is to test and evaluate your conversational agent with real learners and collect data on its usability, functionality, and effectiveness. You can use different methods, such as surveys, interviews, observations, or analytics, to gather feedback from the learners and measure their satisfaction, engagement, learning outcomes, and behavior. You can also use different metrics, such as accuracy, response time, completion rate, or retention rate, to assess the performance of the conversational agent. Based on the results, you can identify the strengths and weaknesses of your conversational agent and make improvements accordingly.

5 Here’s what else to consider

This is a space to share examples, stories, or insights that don’t fit into any of the previous sections. What else would you like to add?