What are we waiting for?
Resources
What are we waiting for?
On 5 November, AI in HE enthusiasts gathered for a mini-symposium focused on Cogniti AI. There were 175 people attending synchronously, and over 1000 registered for the recordings, representing 200 organisations. During the symposium, over 30 speakers, from four countries, shared their journey and success with Cogniti.
What is Cogniti?
The Cogniti website (access it here) explains that “Cogniti is designed to let teachers build custom chatbot agents that can be given specific instructions, and specific resources, to assist student learning in context-sensitive ways.”
Importantly, Cogniti was created by Danny Liu from Sydney University, and as such “is designed by educators, and built by educators, to empower educators.”
Essentially, Cogniti asks you to provide information on the chatbot persona, course content, interaction type/context, and desired outcomes. Cogniti then creates a bot which you can train and trial.
The symposium revealed that Cogniti is most frequently used in two ways: as a general tutoring tool in large, or first-year courses (what one speaker aptly named their stuckbot); or as a niche tool in specialised courses. An impressive array of niche use cases were showcased, such as to improve speaking practice in French class; undertake clinical scenarios; role-play for counselling students with the bot as a patient; analysing case studies; and even a reflective guide.
Key Take-aways?
The biggest take-away from the symposium is that we need to start creating our own chatbots now. What are we waiting for? We do not need to be able to code in order to create a course-specific chatbot. The symposium emphatically demonstrated that platforms like Cogniti have made it (relatively) easy for educators to successfully support student learning by creating tailored, ethical, and pedagogically sound chatbots.
Symposium speakers highlighted that student responses were typically positive, with one speaker’s survey revealing 76% of students strongly agreed that the bot had allowed them to access learning support when and how they needed it. Below, to help you get started, you will find our key take-aways from the day – and, if you’re interested in joining our chatbot creation group, click here.
AI chatbots are like any EdTech tool. They need strategic alignment.
- Orientation: Your chatbot needs to be introduced to students as a complementary tool working alongside teaching and support staff. Many speakers introduced their chatbots in synchronous workshops.
- Support: Students’ usage should be supported. Many speakers explicitly taught students how to prompt, or interact with the tool.
- Integration: Speakers saw direct integration of the chatbot within course materials and activities as key to success. One speaker created a ‘study-buddy-bot’ but didn’t simply leave it as ‘add on support’. Students were asked to make their last conversation prompt a request for a summary of the conversation. This was then submitted in their portfolio.
- Scaffolding: Successful chatbots were fully scaffolded, with both directed and voluntary use of the chatbot at a number of stages in the course.
- Pedagogy: Speakers outlined how they created their bots in alignment with the course’s pedagogy, not as an ‘add on’.
Advice from early adopters:
- Be aware that students may be concerned about using AI. Many were surprised by student hesitancy to engage with their bots.
- Tip: Disclose the ethics of your chatbot and clarify academic integrity.
- Tip: Many took a students-as-partners approach for this very reason.
- Be aware that your agent may not operate perfectly the first time.
- Tip: Consider how you will stage the roll-out, perhaps starting small-scale or low-stakes, so that you can refine your bot.
- Tip: Gather student feedback after every directed session. Cogniti allows bot creators to check usage, responses, and performance.
- Take time to train your bot, as this is key to its success.
- Idea: Enlist students as partners. Have a ‘break the bot’ session with students at the end of one iteration of the course, where students train the bot by asking it every conceivable question – even those we wouldn’t want them to. This allows you to train your bot.
0 Comments