What Zoë REALLY thinks about fiber…

This content is made available under the Creative Commons CC-BY-NC-SA 4.0 License (Attribution-NonCommercial-ShareAlike, View License Deed | View Legal Code).

Overview:

• I’ve long believed that specific instructions beyond “watch this video” (as a minimum), and preferably more engaging activities should accompany video assigned to students.
• Regrettably, most assignments only instruct students to watch or review the video, with no further guidance.
• I consider this expectation unrealistic, as students often watch videos in distracting environments, speed up playback, and multitask.
• I’ve posted several activities that can be created around videos before, but they often require intensive preparation.
• As a solution, I’ve turned to ChatGPT to help develop a simple yet effective way of checking if students have watched a video: a brief multiple-choice content quiz.
• Initially, I encountered difficulties due to model bias and incorrect video summarizations, but I found workarounds for these issues.
• Though ChatGPT can generate reasonably good questions for any video, the generated quizzes still require thorough review.
• One innovation I added was appending time codes to all questions to verify the answers in the video.
• Currently, the model only works on videos with closed captions.
• To conclude, I asked ChatGPT to generate a nicely formatted quiz, complete with MLA or APA formatted citations.
• I firmly believe that with ChatGPT’s help, we can engage students more effectively with video content.
• I’ll be sharing more about this at the upcoming faculty workshop and the Fall ’23 Accelerate Online conference.

Seriously: I thought it would be easy to get Chat-GPT4 Plus to write me a quiz for a YouTube video… but no, here comes the dreaded bias against people who dare think differently… (or is it “think different” aka “the crazy ones, the trouble-makers…”).

>> ————

While it’s often expected that students will take notes while watching an assigned video, as they have done for centuries during lectures, this is an unrealistic and outdated expectation, for several reasons. Unlike the controlled environments of a lecture halls in the limes long gone, where the captive audience had nothing else to do, but listen and scribble, or quiet libraries where the focus was on reading and note-taking, today’s students have many more options, some of which use sophisticated algorithms and multi-million dollar research to command and trap their attention, and they frequently watch videos (read: lectures) in settings where these deliberately programmed distractions compete with one another: listening to music (no, it doesn’t help you study!), checking on Amazon deliveries, browsing social media, or just engaging in household chores: I like doing my laundry while listening (this way you know for sure you won’t fall asleep), and washing the dishes, walking the dog, or commuting are all a real possibility. Additionally, students tend to speed up video playback (anecdote: a few years ago, students at a department which will remain unnamed, refused to use a new video platform because it did not allow speeding things up…), further reducing their comprehension, especially for complex content. So, assuming that students are focused, taking notes, and absorbing 90% of the assigned content is overly optimistic. I’ve previously posted about activities that can be created to increase engagement with assigned videos, but many of these activities require thoughtful advance preparation and, consequently, sometimes a non-trivial commitment of time on the part of the instructor.

I’ve long believed that assigning videos to students is only beneficial if they’re also given specific instructions on what to do with the content they are watching and where to focus their attention. Long time ago (2020) I wrote an assertively titled post: An unexamined video is not worth the time spent watching… I still believe that. The first recommendation in that older post, “multiple-choice activity,” recommended a quick content-check quiz as a simple, minimalistic yet effective way to encourage at least some engagement with the video content (my recommendations for implementation, timing, and NOT calling it a quiz still stands, and you can read about it here). But even that, unfortunately, requires some prep time (not a lot for a 10 min. video, but quite a bit for a 45 min. lecture!); and so most video watching assignments simply ask students to view the video, and that’s where the instructions end. It’s often assumed that students will take notes as they watch, similar to how they might during lectures, or it’s assumed that they will watch the video and this will be the basis for their later discussion posts. And so, I thought artificial intelligence could help me here (and thousands of busy instructors who inadvertently assigned videos and now discover that few students really watch them) and generate a simple content quiz based on a YouTube video.

As of July 2023, the basic, free version of ChatGPT-3 cannot access online videos (although some competition can), but the paid version, ChatGPT-4 Plus, allows the use of plugins to scour the web, including YouTube auto-generated transcripts, for relevant information. My goal was to prompt ChatGPT to “watch a video” (actually, really to simply analyze the transcript) and then generate first a true/false, and then a multiple-choice quiz, with 10-15 questions, to verify students’ comprehension and promote at least a minimal direct engagement with video content. This turned out to be more complex than I expected, and I encountered several issues during my early attempts.

One of the biggest problems was the AI’s inherent bias, as it seemed to consistently and persistently (despite clear instructions to the contrary) favor online majority consensus over the specific content of the video.This is not a surprise: we know that any AI LLM-based system will reflect the views of the material on which it was trained (despite ChatGPT-4’s self-declaration of not having or favoring any views).

To test this, I deliberately selected a video that would epitomize views that are sharply at odds with mainstream media consensus – a video on dietary fiber by Zoë Harcombe PhD, a Cambridge U/ educated independent scholar, a popular public-debate figure active in the fast-growing field of commercial determinants of health. This field explores how commercial entities, especially very large international corporations in chemical / food and pharmaceutical industries strategically influence public health advice for profit through political activity (lobbying, public image campaigns, support of favorable research, and funding of major research efforts and consortia). Perhaps the best general-audience overview of this field and research is Chris van Tulleken Zoë Harcombe’s views – while grounded in solid science and meticulous analysis of published research – often address some of these tactics, dissecting articles, and exposing conflicts of interests in research that seems to promote ideas that benefit the corporate sponsors. I wanted to see if the AI LLM model would reflect her perspective rather than the mainstream media views driven by corporate funding. Initially, ChatGPT reflected widespread “consensus” opinions, so I had to make numerous attempts to fine-tune the prompt to focus on the video’s content and exclude references to general knowledge or prevailing opinions. After a trial-and-error process, I was able to guide ChatGPT to generate questions more closely aligned with the specific content of the video.

Other issues included ChatGPT often targeting the wrong video and focusing mostly on the first few minutes of the video. By providing the video’s URL and title and instructing ChatGPT to spread the questions evenly across the video’s length, I could address these issues. Moreover, the answer pattern was initially too predictable. But by requesting that at least 40% of the questions should be true and another at least 40% should be false, I could make the quiz more varied and challenging.

While it’s not perfect (it still cannot easily handle videos longer than 20 minutes, although I’m working on a work-around), ChatGPT is now able to generate a reasonably good first draft of a quiz. Still, I wouldn’t use such a quiz without first verifying that the questions and answers align with the video content. To simplify this, I asked ChatGPT to append timestamps to all questions, indicating where  in the video the correct answer can be verified.  However, one important limitation is that this only works on videos that have closed captions. Also, to make the quiz more presentable, I asked ChatGPT to include formatted citations, video views, the title, author, and link, followed by an automatically generated answer key.

There’s no longer an excuse to assign a video with simple instructions like “watch the video” or “review the video.” With minimal effort and time, ChatGPT can generate a draft of a true/false or multiple-choice quiz. With some tweaking and occasional checks for answer correctness, we can ensure students engage more deeply with the content.

If you are one of the faculty my team works you will have the opportunity to see an example quiz, and understand the process, if you join me at the August 24 faculty workshop, or sign-up for the Teaching with Multimedia Fellowship. Or, if you are at another University, perhaps somewhere far from here, you can still can join me and my departmental colleague at the Fall ’23 Accelerate Online conference (October 3-5, 2023) where we will be sharing this, and more ChatGPT4-related insights. 

NOTE: The following text is AI-Generated

Video Information:
– Title: What about fibre?
– Link: [https://youtu.be/tRQ2ciJ1ncQ](https://youtu.be/tRQ2ciJ1ncQ)
– Author: Dr Zoe Harcombe PhD
– Views: 1,000,000+ (NOTE: This is incorrect – actually it’s 26.3K subscribers and 35K views) 

Key Points:
1. There is no randomized control trial or Cochrane evidence to support the claim that fiber is essential for health (0:00-3:00).
2. The benefits of fiber are often associated with the benefits of the foods that contain fiber, not the fiber itself (3:00-6:00).
3. The consumption of fiber can have negative effects, such as rushing nutrients through the body and potentially impacting mineral absorption (6:00-9:00).
4. High-fiber diets can improve gut flora, but so can other factors like natural birth, breastfeeding, and avoiding antibiotics (9:00-12:00).
5. High-fiber diets are often compared to diets high in refined carbohydrates, not low-carb diets (12:00-15:00).
6. A low-carb diet is not necessarily a low-fiber diet (15:00-18:00).

>> NOTE: despite my efforts, content from mins. 18-35 is not included…

Summary:
In the video “What about fibre?” by Dr Zoe Harcombe PhD, she challenges the common belief that fiber is essential for health. She points out that there is no randomized control trial or Cochrane evidence to support this claim. Instead, she suggests that the benefits often associated with fiber are actually benefits of the foods that contain fiber, not the fiber itself. She also highlights potential negative effects of fiber consumption, such as rushing nutrients through the body and potentially impacting mineral absorption. While acknowledging that high-fiber diets can improve gut flora, she points out that other factors like natural birth, breastfeeding, and avoiding antibiotics can also have this effect. She criticizes the common comparison of high-fiber diets to diets high in refined carbohydrates, arguing that a more relevant comparison would be to low-carb diets. In fact, she concludes that a low-carb diet is not necessarily a low-fiber diet.

True/False Quiz:
1. Fiber is essential for health. (False, 0:00-3:00)
2. The benefits of fiber are solely due to the fiber itself. (False, 3:00-6:00)
3. Fiber consumption can negatively impact mineral absorption. (True, 6:00-9:00)
4. Only high-fiber diets can improve gut flora. (False, 9:00-12:00)
5. High-fiber diets are often compared to low-carb diets in studies. (False, 12:00-15:00)
6. A low-carb diet is necessarily a low-fiber diet. (False, 15:00-18:00)
7. There is robust evidence to support the claim that 14 grams of fiber per day per thousand calories is beneficial. (False, 0:00-3:00)
8. Foods that contain fiber are generally healthier than those that do not. (True, 3:00-6:00)
9. Fiber can slow down the digestion process. (True, 6:00-9:00)
10. All diets high in fiber are also high in carbohydrates. (False, 15:00-18:00)

Multiple Choice Version:

1. According to the video, is fiber essential for health? (0:00-3:00)
    – A. Yes, it is essential for everyone.
    – B. Yes, but only for some people.
    – C. No, it is not essential.
    – D. The video does not mention this. 
    – Correct Answer: C

2. According to the video, are the benefits of fiber solely due to the fiber itself? (3:00-6:00)
    – A. Yes, the benefits are solely due to the fiber.
    – B. No, the benefits are due to the foods that contain fiber.
    – C. The benefits are due to both the fiber and the foods that contain it.
    – D. The video does not mention this.
    – Correct Answer: B

3. According to the video, can fiber consumption negatively impact mineral absorption? (6:00-9:00)
    – A. Yes, it can negatively impact mineral absorption.
    – B. No, it has no impact on mineral absorption.
    – C. It can improve mineral absorption.
    – D. The video does not mention this.
    – Correct Answer: A

4. According to the video, can only high-fiber diets improve gut flora? (9:00-12:00)
    – A. Yes, only high-fiber diets can improve gut flora.
    – B. No, other factors like natural birth and breastfeeding can also improve gut flora.
    – C. No, only low-fiber diets can improve gut flora.
    – D. The video does not mention this.
    – Correct Answer: B

5. According to the video, are high-fiber diets often compared to low-carb diets in studies? (12:00-15:00)
    – A. Yes, they are often compared to low-carb diets.
    – B. No, they are often compared to diets high in refined carbohydrates.
    – C. They are compared to both low-carb diets and diets high in refined carbohydrates.
    – D. The video does not mention this.
    – Correct Answer: B

6. According to the video, is a low-carb diet necessarily a low-fiber diet? (15:00-18:00)
    – A. Yes, a low-carb diet is always a low-fiber diet.
    – B. No, a low-carb diet is not necessarily a low-fiber diet.
    – C. It depends on the specific foods in the diet.
    – D. The video does not mention this.
    – Correct Answer: B

7. According to the video, is there robust evidence to support the claim that 14 grams of fiber per day per thousand calories is beneficial? (0:00-3:00)
    – A. Yes, there is robust evidence to support this claim.
    – B. No, there is not robust evidence to support this claim.
    – C. The evidence is mixed.
    – D. The video does not mention this.
    – Correct Answer: B

8. According to the video, are foods that contain fiber generally healthier than those that do not? (3:00-6:00)
    – A. Yes, foods that contain fiber are generally healthier.
    – B. No, foods that do not contain fiber can be just as healthy.
    – C. It depends on the specific food.
    – D. The video does not mention this.
    – Correct Answer: A

9. According to the video, can fiber slow down the digestion process? (6:00-9:00)
– A. Yes, fiber can slow down the digestion process.
    – B. No, fiber speeds up the digestion process.
    – C. Fiber has no impact on the digestion process.
    – D. The video does not mention this.
    – Correct Answer: A

10. According to the video, are all diets high in fiber also high in carbohydrates? (15:00-18:00)
    – A. Yes, all high-fiber diets are also high in carbohydrates.
    – B. No, not all high-fiber diets are high in carbohydrates.
    – C. It depends on the specific foods in the diet.
    – D. The video does not mention this.
    – Correct Answer: B

APPENDIX
Brief on Commercial Determinants of Health

NOTE: The content below has been generated in response to human-generated, carefully crafted prompts by ChatGPT-4 Plus (with plugins enabled)

Overview:

Definition:
The commercial determinants of health (CDoH) refer to strategies and approaches used by the corporations to promote products and choices that are beneficial to their financial and strategic interests, but often simultaneously detrimental to health. These determinants include advertising, marketing, product design and placement, pricing, political lobbying, and financial support for research that favors financially preferrable outcomes.

History:
The concept of commercial determinants of health emerged in the late 20th century, as public health researchers began to identify and understand the role of industry and commerce in shaping health outcomes. This was initially recognized in the context of tobacco and alcohol, but has since expanded to include a wide range of industries.

Related Terms:
Social Determinants of Health (SDOH): These are conditions in the environments in which people are born, live, learn, work, play, worship, and age that affect a wide range of health, functioning, and quality-of-life outcomes and risks.
Health Equity: This refers to the attainment of the highest level of health for all people, achieved by ensuring equal opportunities for everyone to achieve their full health potential.
Health Promotion: This is the process of enabling people to increase control over, and to improve, their health.

Resources:
Text: Kickbusch, I., Allen, L., & Franz, C. (2016). The commercial determinants of health. The Lancet Global Health, 4(12), e895-e896. Link
Text: Moodie, R., Stuckler, D., Monteiro, C., Sheron, N., Neal, B., Thamarangsi, T., … & Casswell, S. (2013). Profits and pandemics: prevention of harmful effects of tobacco, alcohol, and ultra-processed food and drink industries. The Lancet, 381(9867), 670-679. Link
Video: “Commercial Determinants of Health” – A lecture by Prof. Jeff Collin, University of Edinburgh. REMOVED
Audio: “The Food Industry’s ‘Predatory’ Health Tactics” – A podcast episode by BBC’s The Food Chain. REMOVED
Text: World Health Organization. (2019). Commercial determinants of health: understanding and influencing the commercial determinants. Link

Basic Intro Article:

APA Citation: Kickbusch, I., Allen, L., & Franz, C. (2016). The commercial determinants of health. The Lancet Global Health, 4(12), e895-e896. https://doi.org/10.1016/S2214-109X(16)30217-0
MLA Citation: Kickbusch, Ilona, Allen, Luke, and Franz, Christian. “The commercial determinants of health.” The Lancet Global Health, vol. 4, no. 12, 2016, pp. e895-e896, doi:10.1016/S2214-109X(16)30217-0.
Key Points:
> The article discusses the commercial determinants of health, focusing on how private sector strategies promote products and choices detrimental to health.
> It highlights the conflict between shareholder value maximisation and population health.
> The authors define the commercial determinants of health as strategies and approaches used by the private sector to promote products and choices that are detrimental to health.
> The article discusses how corporate influence is exerted through marketing, lobbying, corporate social responsibility strategies, and extensive supply chains.
> The authors argue that the rise of non-communicable diseases is a manifestation of a global economic system that prioritises wealth creation over health creation.
Summary: The article “The commercial determinants of health” by Ilona Kickbusch, Luke Allen, and Christian Franz, published in The Lancet Global Health in 2016, explores the impact of commercial activities on public health. The authors argue that the strategies and approaches used by the private sector often promote products and choices that are detrimental to health, leading to a rise in non-communicable diseases. They highlight the conflict between the drive for shareholder value maximisation and the need for population health. The authors also discuss the various channels through which corporations exert their influence, including marketing, lobbying, corporate social responsibility strategies, and extensive supply chains. They conclude by stating that the current global economic system, which prioritises wealth creation over health creation, is a significant contributor to the rise of non-communicable diseases.

AI-Generated T/F Quiz with answer key
1 – The article argues that efforts to prevent non-communicable diseases are in line with the business interests of powerful economic operators. (True/False)
2 – The authors suggest that the power of the corporate sector has always been critically analyzed in public health, especially in the field of tobacco. (True/False)
3 – The authors propose the unification of various domains of study under the banner of the commercial determinants of health. (True/False)
4 – The authors define the commercial determinants of health as strategies and approaches used by the private sector to promote products and choices that are beneficial to health. (True/False)
5 – The authors argue that corporate influence is exerted through marketing, lobbying, corporate social responsibility strategies, and limited supply chains. (True/False)
6 – The article suggests that the rise of non-communicable diseases is a manifestation of a global economic system that currently prioritises health creation over wealth creation. (True/False)
7 – The authors believe that there is no quick fix for misaligned incentives, uncorrected externalities, and widespread cognitive dissonance. (True/False)
8 – The authors suggest that the commercial determinants of health framework presented in the article is intended to be exhaustive. (True/False)
9 – The authors argue that the current piecemeal efforts to combat the commercial drivers of non-communicable diseases are sufficient. (True/False)
10 – The authors declare that they have competing interests. (True/False)
Answer Key (starting line of the paragraph included): 1 – False – “WHO Director-General Margaret Chan…” 2 – True – “There has always been critical…” 3 – True – “These domains of study share…” 4 – False – “We define the commercial determinants…” 5 – “Corporate influence is exerted through…” 6 – True – “The rise of non-communicable diseases…” 7 – True – “There is no quick fix…” 8 – False – “The commercial determinants of health…” 9 – False – “In the vernacular of our…” 10 – False – “We declare no competing interests…”