For organisations who are serious about embracing AI, training their staff to have a basic AI competency is the first step. Hence, NUS has mandated this Basic Artificial Intelligence Competency Course (AICC) for all staff.
AICC is a five-week programme that focuses on training NUS staff (all departments and disciplines) to be conversant in AI, identify opportunities to apply AI in the workplace and have the ability to initiate AI projects. I recently completed AICC on 17 May 2023 and am sharing my AICC reflections for this month’s blog entry.
What I liked
1. Real world examples
Our instructor, Ms Cui Wei shared a mix of global case studies and local real world examples. Even for the local examples, they chose one (NUS autonomous shuttle, 2019) that is related to our own institution. By contextualising it to our learners’ profiles, it made it even closer for us to relate. Nice.
2. Clear instructions, content segmented into manageable chunks weekly
Clear instructions on the structure of the course, the assessment criteria (x4 weekly quizzes & x1 final group project presentation) and logical flow of content categorised and chunked week by week. Learners know in advance what's in store each week. Even when our instructor had a relief instructor stand in, there was no disruption to the teaching or learning.
3. Project-based group work
The final assignment was a group project that challenged the team's understanding of business data (including preparation, modelling and evaluation), the societal impact of our use of AI to the job market, and potential discrimination or bias.
I had the good fortune of partnering with four great team mates. Although from different departments, all were committed learners and contributed actively to the group project. We eventually decided to work on Harry’s real case scenario of sourcing for relief instructors for his medical school. Throughout the weeks, hearing my team members’ different perspectives and ideas was like having a new lens that opens up my eyes on new approaches to problem solving.
On the day of assessment, we cleared the presentation smoothly. Next is to wait and see if those plans can be put into action. Regardless, we now know how to speak the AI-related technical jargon should other opportunities (to get involved with an AI-related project) present itself.
4. Combating Ebbinghaus’ Forgetting curve & Using Learners data
At the beginning of each week’s class, our instructor will do two things:
recap last week’s key concepts before moving onto the new topic. This quick recap allows learners to refresh their understanding and mitigate the forgetting curve. Especially helpful for classmates who might be busy that last week and unable to catch up on revision.
analyse students’ learning data (gathered from students’ quiz scores) to identify which questions students had more difficulty with and discuss those 2 or 3 in class. This targeted approach only takes about 10-15 mins but allows most students’ to clarify issues and see how we measure up among our peers. Using peer pressure to keep us focused.
While I thoroughly enjoyed the course, there will always be areas that can be enhanced. Here are some ideas I have that may be useful for future runs.
What can be improved
1. Introduce naive tasks
Have students work on naive tasks, or learning activities prior to class. Currently, pre-readings (homework) are assigned in the form of pre-recorded lectures. However, this is largely one-way and has minimal effect on Social-presence. Perhaps setting up a Padlet or Miroboard for students to share some insights, some learning activities can be added as pre-class as well as during class to engage the students.
2. Use Audience Response Systems
During the class itself, for the theory portion, it was mainly didactic. It was not as slow because the course content was already pre-recorded and made available to students. However, we could have used tools like Polleverywhere or Kahoot to get the class more involved during the lesson.
3. Enhance pre-recorded lectures
I felt that there was excessive front loading on use of RapidMiner Studio. To learn this software*, we were instructed to install the software, watch x7 pre-recorded tutorial videos and practise with the exercise files provided. The non-software topics were slightly better. However, compared to those examples I saw on Linkedin Learning, these pre-recorded videos could be enhanced by using annotations or simple visual effects like zoom-and-pan as well as a more informal style of presenting.
*While the software is lauded for its graphical user interface, personally I did not find it that easy-to-use. Nonetheless, I was glad to be introduced to a visual AI software tool that was able to quickly generate tables, plot graphs instantly and generate a decision tree.
4. Use the LMS built-in forums
Although we had students committed to the course for five consecutive weeks, the forums were hardly used. There was only one post, and one reply. If managed well, this could help learners communicate and link up with one another, building up the social presence along the way.
In conclusion
In the daily grind of work, it may be difficult to generate new insights. Similar to how ONL designed their course, I liked how AICC also designed its course centred around a central theme, grouped learners into specific teams, and set problems that could only be achieved through teamwork. Having a common goal to work towards but only achievable through learners’ collaboration is a powerful motivator. Well played Ms Cui Wei & AICC admin team!
AI is here to stay. Understanding how AI works, what are its current limitations (bias, adversarial AI) and its seemingly incredible potential, can help us better adapt the way we work in ways that leverage AI's strengths, while staying clear of its pitfalls.
Benedict Chia
27 May 2023
Comments