Week in Review | 29-23
Welcome to my latest experiment, a weekly blog post series where I share what I’ve been listening to on Soundcloud, what articles I’ve been reading or am reading, and which podcasts have caught my attention this week. I’m excited to share my latest discoveries with you and hope you’ll find them as interesting and enjoyable as I have. Here are my latest picks and what’s been on my radar this week!
What I’m Listening to
What I’m Reading
Ginns, P., Martin, A. J., & Marsh, H. W. (2013). Designing instructional text in a conversational style: A meta-analysis. Educational Psychology Review, 25, 445-472.
A study was done to see if using a conversational writing style in teaching materials would help people learn better. The results showed that materials written in a conversational style were more effective in helping people remember and use what they learned. People also liked the conversational style more and found it easier to understand. However, the effects on interest and perceived learning assistance were smaller and not always reliable. Overall, this study suggests that conversational style can be helpful in teaching, but there may be limits to how effective it is.
Veletsianos, G., & Russell, G. S. (2014). Pedagogical agents. Handbook of research on educational communications and technology, 759-769.
This chapter looks at research on pedagogical agents from 2005 to 2011. Researchers think these agents can be helpful in teaching by simulating real-life situations, encouraging engagement, motivation, and responsibility, and improving learning. However, there is mixed evidence to support these claims. Previous studies focused on cognitive issues, but there is now more investigation into sociocultural factors.
This study found that ChatGPT’s accuracy in answering questions varies depending on the subject and difficulty level, ranging from 61% to 95%, with an average of 83.5%. There is still room for improvement, particularly in handling questions with “None” answers. It is noted that question construction plays an important role (just like questions for humans would be). ChatGPT may be helpful to MOOC students, but it’s not a substitute for human instructors. It has the potential to enhance education by providing personalized and interactive learning experiences. However, it should be balanced with other learning strategies that encourage collaboration and engagement with peers.
This study shows that using tangible rewards, which can be redeemed through intangible rewards, is more effective in stimulating student engagement and learning performance in gamified education than relying solely on intangible rewards such as virtual points and badges. The tangible rewards group outperformed the intangible rewards group in terms of intrinsic motivation, behavioral engagement, cognitive engagement, and learning performance. Practitioners should consider using tangible rewards and linking them to a standard of performance, as well as setting specific and moderately challenging redemption goals to enhance the effectiveness of gamified learning.
Khanmigo is great but NOT ready to tutor student via Richard Tong
“I played with it for several hours and here is the TLDR of my experience:
In the “Tutor me” mode, Khanmigo is very good at guiding the tutoring process by providing tailored scaffolding support, asking probing questions, prompting critical thinking and rendering math formula and solutions, but it had one fatal issue — its answers are not trustworthy. Even for basic elementary school math, I have encountered incorrect answers. For more advanced math or physics problem that requires more computation, Khanmigo is even worse.”
Read the full article for Richard’s breakdown. This runs along the same lines we’ve seen elsewhere around LLMs as assistive tools for learning. It’s worth distinguishing between a process coach vs. a definitive answer, and I think that’s a note-worthy product design note as more people build tools using LLMs. Not to say human teachers are infallible either – I used to teach high school math and had been corrected by my students more than once, which I think is a demonstration of the potential power of group learning – but if we are placing bots in place of human teachers, with a similar power dynamic in place (trust, authority), then we need to consider fallibility.
331. Boris Johnson’s WhatsApps… – Smashing Security
RFK Jr. and The Rise of the Antivax Movement – Maintenance Phase
The CANADALAND Guide to the Navigator – Canadaland
Instructional Design Models Dr. Tonia Dousay (#20) – StyleLearn Origins
I’ve been following Tonia’s work for some time, and often reference the open textbook chapter she mentions here, as well as my freshly minted copy of the Instructional Design Models Survey book. This is a great conversation about ID models, their origins, their purpose, and much more.