On 6 November 2026, AICET welcomed a delegation from the Centre for Holistic Teaching and Learning at Hong Kong Baptist University (HKBU). The visit was co-hosted by AICET Director Prof. Ben Leong and Deputy Director (Research and Pedagogy) Prof. Soo Yuen Jien, and centred on ScholAIstic, AICET’s platform for educators to design AI-enabled learning activities.
The session covered the platform’s features, how NUS educators are using it in their teaching, and what is coming next.
Design, monitor, evaluate
ScholAIstic lets educators design, monitor and evaluate AI-enabled learning activities. To date, AICET has worked on 23 projects with NUS educators, with another 14 under way, across 14 faculties and schools and more than 3,000 students. The platform is also used by 13 MOE schools and polytechnics, and by the SAF, SPF, the Judiciary and the Legal Aid Bureau.
Educators choose an activity format, decide how the AI supports the task through prompts, models and reference materials, and set the rules: attempts, time limits, how an activity ends, and the order in which activities unlock. Conversations between students and the AI are saved automatically, and educators can view them live.
Across the projects co-designed with NUS educators, five pedagogical approaches have emerged: simulation-based practice, Socratic tutoring, inquiry-based learning, metacognitive scaffolding and collaborative learning.
How NUS educators are using ScholAIstic
Examples ranged from single-character assignments to activities spanning a whole course. Business students negotiate a job offer with AI recruiters of differing communication styles, Dentistry students interpret radiographs with an AI guide, and Pharmacy students take histories from an AI patient before justifying their reasoning to an AI preceptor.
Newer designs link activities together. In a Business School IPO assignment, an AI mentor reads students’ interviews with an AI CEO and CFO before helping them refine their pitch. In a Computing leadership course, an AI “Inner Voice” prompts students to reflect mid-conversation, and is now the subject of a controlled study.
Feedback for students, insight for educators
Educators write their own evaluation criteria, using their rubric or framework, and decide when students see the feedback: immediately, or after a tutor has reviewed it. Beyond feedback on each attempt, ScholAIstic supports comparative feedback. In the leadership course, students hold the same conversation with an AI manager twice, and a second report compares the two attempts and checks which advice the student acted on.
On the educator’s side, live transcripts show every conversation as it happens, and a data assistant answers questions about performance across the whole class.
Before an activity reaches students, educators can test it. They write an AI persona, such as a novice interviewer, and let it talk to the chatbot to check that the chatbot stays in role. The same test can run side by side on two models or two versions of a prompt, and the evaluation can be checked on sample transcripts before any student sees it.
What’s next
The next phase of ScholAIstic focuses on human–AI collaboration through multimodal and multi-user design. A new document workspace places an AI chat beside the student’s writing, with AI actions that educators set up and suggested changes students can accept or reject. Keystroke logging lets educators replay how a piece of writing took shape.
Educators can also set how involved the AI is, along a spectrum from no AI, to an inquiry guide that only asks questions, to an adviser that suggests changes, to a co-creator that contributes ideas of its own. Multi-user chat rooms are also in development, opening up activities such as multi-party negotiations, AI-facilitated group discussions and student teams interviewing an AI client.