Thursday 29 June 2023

Myths and realities about AI


Myths and realities about AI

Mildred López


Yesterday, as I was preparing my daily reflection about Artificial Intelligence (AI), I did a simple search in Google for the prompt “artificial intelligence is….”. I’ll share with you a screenshot of the 10 first predictions that the search bar got me: 


Figure 1: Screenshot of Google search


These represent common myths or assumptions about AI. I would love for us to reflect together; therefore, I’ll share a video for each and my comments:


1. Artificial Intelligence is not prone to any bias or discrimination.

The reality is that it is prone to bias. The first step is to recognize them and build systems to address them. It is imperative that we understand how algorithms work and critically use the information it provides. 

See Jabril’s collab with CrashCourse shares 5 types of bias:

- Data reflects existing biases.

- Unbalanced classes in training data.

- Data doesn’t capture the right value.

- Data amplified by a feedback loop.

- Malicious data attack or manipulation.

https://www.youtube.com/watch?v=gV0_raKR2UQ


2. Artificial Intelligence is dangerous. 

Define dangerous. I want to share with you the analysis of Joanna Stern in the Wall Street Journal. She used several challenges to see if AI with pass as a human. In several cases, it already passed:

https://www.youtube.com/watch?v=t52Bi-ZUZjA


3. Artificial Intelligence is going to take over our homes.

I am convinced it will get there, especially in monotonous tasks that can be automated. Tilman Wolff and Ranga Yogeshwar documentary for DW Documentary explains different applications, its pros, and cons:

https://www.youtube.com/watch?v=s0dMTAQM4cw

 

4. Artificial Intelligence is about.

Listen to the excerpts of Stuart Rusell on Radio Davos:

https://www.youtube.com/watch?v=RzkD_rTEBYs


5. Artificial Intelligence is a threat to humanity debate.

Beyond sci-fi movies, we need ground rules to work. Previous development rules for AI are no longer in place. Mo Gawdat (former Chief Business Officer for Google X) discusses with Piers Morgan the speed at which AI is learning. He mentioned 3 boundaries that they had for AI: 

- don’t put it on the open internet until the control problem is solved, 

- don’t teach it to code because that makes it self-develop,

- don’t have other agents prompting it.

I can’t forget one line of the interview where they mentioned, “I am not concerned of the machines…I am concerned of humans with AI in their hands”. You can see the full interview here: 

https://www.youtube.com/watch?v=oxRZqzth9r4

We need to focus not only on the rules for development but also on the implementation rules of these platforms. 


6. Artificial Intelligence is about mcq.

This prompt represents questions, multiple-choice questions about AI. It is a great opportunity to test our understanding of the topic. I’ll like to share a WIRED video of Gary Marcus, an expert on AI, answering questions from Twitter:

https://www.youtube.com/watch?v=Puo3VkPkNZ4


7. Artificial Intelligence is associated with computers of which generation.

The concept of generations of computers is explained briefly by Simply Coding at: 

https://www.youtube.com/watch?v=NqgpZ_v4Ne8


8. Artificial Intelligence issues.

The Center for Humane Technology presents Tristan Harris and Aza Raskin, as they discuss the dilemma of AI development. See the full video at:

https://www.youtube.com/watch?v=xoVJKj8lcNQ


9. Artificial Intelligence is good or bad.

We cannot categorize it as good or bad. AI will work to do what we ask it to do, therefore, it depends on who is using it. See Janelle Shane TED conference on the danger of AI is weird than you think. She presents several examples for discussion.

https://www.youtube.com/watch?v=OhCzX0iLnOc


10. Artificial Intelligence is defined as.

I enjoyed watching this video by Duke University; it leverages our understanding of pop culture, such as movies and series, to explain a complex concept in a relatable way:

https://www.youtube.com/watch?v=c0m6yaGlZh4


What other myths and realities do you know about AI?

 



Explaining Artificial Intelligence in 5-levels of complexity


Explaining Artificial Intelligence in 5-levels of complexity

Mildred López (accompanied by Chatgpt)


I love the WIRED series that ask an expert to explain a complex concept in 5-levels of complexity; these allow me to analyze a concept from different perspectives. Here is an example: 

https://www.youtube.com/watch?v=VTvJOb2hJ6k


Today, I’ll attempt to do the same and explain Artificial Intelligence (AI) in 5 different ways. To do so, I received the help of Chatgpt. These prompts were used on June 29 at 5 pm at: 

https://chat.openai.com/


For each, I'll share with you a shortened version of the responses for each level of complexity and the prompts that I used:


1. Artificial intelligence to a 10-year-old

Response of Chatgpt

Imagine you have a robot friend who can do things just like you, but it doesn't have a brain like yours. AI is a way to give that robot a brain so it can learn and make decisions by itself, just like humans do.

AI uses special computer programs to teach the robot how to do things. The robot learns from a lot of information, just like you learn from books and experiences. It looks at all the information, finds patterns, and uses those patterns to figure out how to solve problems or make choices.

For example, let's say the robot wants to recognize pictures of cats. You show it many pictures of cats and tell it, "These are cats." The robot learns what features or clues make something a cat, like pointy ears, whiskers, and a tail. Then, when you show it a new picture, the robot can decide if it's a cat or not based on what it has learned.

Prompt:

Explain artificial intelligence to a 10-year-old


2. Artificial intelligence to a teen

Response of Chatgpt:

Artificial intelligence is an exciting and rapidly advancing field of technology that focuses on creating computer systems that can perform tasks that usually require human intelligence. It's like teaching computers to think and learn just like humans do.

AI is all about building algorithms and models that enable machines to analyze data, recognize patterns, and make decisions. It involves training these machines on lots of information so that they can learn from it and become smarter over time.

One fascinating aspect of AI is machine learning. Machine learning algorithms allow computers to learn from examples and experiences, instead of being explicitly programmed for every specific task. Just like how you learn from practice and experience, AI systems get better at what they do with more data and training.

Prompt:

Explain artificial intelligence to a teen


3. Artificial intelligence to a college student

Response of Chatgpt:

Artificial intelligence is a fascinating field that focuses on creating intelligent machines capable of performing tasks that typically require human intelligence. It encompasses a wide range of techniques and technologies that aim to mimic or simulate human cognitive abilities using computers.

AI involves the development of algorithms and models that enable machines to process and understand information, learn from data, reason and make decisions, and interact with the world in meaningful ways. The goal is to create systems that can perceive, comprehend, and act in complex and dynamic environments.

Machine learning is a fundamental aspect of AI. It involves training computers to learn from data, allowing them to improve their performance on specific tasks without being explicitly programmed. Machine learning algorithms analyze patterns in data, identify relationships, and make predictions or classifications.

Prompt:

Explain artificial intelligence to a college student


4. Artificial intelligence to a college student

Response of Chatgpt:

Artificial intelligence is a multidisciplinary field that focuses on developing intelligent systems that can perform tasks requiring human-like intelligence. It encompasses various techniques, algorithms, and technologies that enable machines to perceive, reason, learn, and make decisions in complex and dynamic environments.

AI involves the creation of models and algorithms that allow computers to process and understand information, learn from data, and perform tasks autonomously. It aims to replicate human cognitive processes, such as perception, problem-solving, and decision-making, using computational methods.

Machine learning is a fundamental aspect of AI. It involves training computers to learn from data and improve their performance on specific tasks without being explicitly programmed. Machine learning algorithms can discover patterns, extract meaningful insights, and make predictions or decisions based on the learned information.

Prompt:

Explain artificial intelligence to a grad student


5. Artificial intelligence to an expert

Response of Chatgpt:

Artificial intelligence is a multidisciplinary field that focuses on creating intelligent systems capable of performing tasks that typically require human intelligence. It encompasses various subfields, including machine learning, deep learning, natural language processing, computer vision, and more.

At its core, AI is concerned with developing algorithms and models that allow machines to process and analyze data, learn from it, and make intelligent decisions or predictions. The goal is to enable machines to mimic or surpass human cognitive abilities, such as perception, reasoning, problem-solving, and decision-making.

Machine learning is a key component of AI, where algorithms learn from data and automatically improve their performance without being explicitly programmed. Supervised learning algorithms learn from labeled data to make predictions or classifications. Unsupervised learning algorithms find patterns in unlabeled data and discover hidden structures. Reinforcement learning involves training an agent to make decisions based on feedback from its environment.

Prompt:

Explain artificial intelligence to an expert on AI


Can you spot the differences? Do you think Chatgpt did a good job layering different styles to present the information? 

 



Curating Literature and Commentary on AI in HPE

Curating Literature and Commentary on AI in HPE

Poh-Sun Goh


Will be progressively curating, and selectively adding comments as part of an ongoing process to curate literature and commentary on AI in HPE.


Shankar, P. Ravi. Artificial Intelligence in Health Professions Education. Archives of Medicine and Health Sciences 10(2):p 256-261, Jul–Dec 2022. | DOI: 10.4103/amhs.amhs_234_22

https://journals.lww.com/armh/Fulltext/2022/10020/Artificial_Intelligence_in_Health_Professions.20.aspx


Lomis, K., Jeffries, P., Palatta, A., Sage, M., Sheikh, J., Sheperis, C., & Whelan, A. (2021). Artificial Intelligence for Health Professions Educators. NAM perspectives, 2021, 10.31478/202109a. https://doi.org/10.31478/202109a


Gray K, Slavotinek J, Dimaguila GL, Choo D. Artificial Intelligence Education for the Health Workforce: Expert Survey of Approaches and Needs. JMIR Med Educ 2022;8(2):e35223. doi: 10.2196/35223

https://mededu.jmir.org/2022/2/e35223


Jacob Krive and others, Grounded in reality: artificial intelligence in medical education, JAMIA Open, Volume 6, Issue 2, July 2023, ooad037, https://doi.org/10.1093/jamiaopen/ooad037


Dave, M., Patel, N. Artificial intelligence in healthcare and education. Br Dent J 234, 761–764 (2023). https://doi.org/10.1038/s41415-023-5845-2 ".. This article has been entirely written by the artificial intelligence (AI) software ChatGPT."


Ellaway, R.H., Tolsgaard, M. Artificial scholarship: LLMs in health professions education research. Adv in Health Sci Educ (2023). https://doi.org/10.1007/s10459-023-10257-4


Arbelaez Ossa, Laura & Rost, Michael & Lorenzini, Giorgia & Shaw, David & Elger, Bernice. (2023). A smarter perspective: Learning with and from AI-cases. Artificial Intelligence in Medicine. 135. 102458. 10.1016/j.artmed.2022.102458.  https://www.sciencedirect.com/science/article/pii/S093336572200210X

AI Describing AI - Being Meta About AI

 AI Describing AI - Being Meta About AI

Sol Roberts-Lieb

29 June 2023, 9:40am, Central US Time


As we know, artificial intelligence has the potential of changing our lives and education dramatically. The question is not if we should use it or not, but how and to what extent. How can we help our faculty, staff, and students understand the pros and cons of AI and what it means to them? 

Food for thought.  How does the use of AI extend beyond creating text to solve problems or provide answers? Where and how do we know what truth is when AI is used. 

The first set of images below was developed through Adobe Firefly. This is a new tool to turn text into art. 

Prompt: "Artificial intelligence in health professions education"

Images: 


Questions: What does this mean to you? Does this encapsulate the prompt? Is this a work that I should get credit for even if I cite using Adobe Firefly?


The second image below was also developed by using Adobe Firefly. It is a description of my compared to an actual picture of me. 

Prompt: "a balding 46 year old Caucasian male weighing 300 pounds, six foot two inches in height with black hair, trimmed beard, and glasses. He is wearing a blue plaid shirt with khaki shorts. he is of Russian and Romanian descent"

Images: 



Questions: While it isn't an exact match, it's close. How can we know what is AI and what isn't?


Share your thoughts with us. How doe these examples help us understand the use of AI in Education? 







 

Wednesday 28 June 2023

Augmenting Human Intelligence - engaging in a broader discussion

Augmenting Human Intelligence - engaging in a broader discussion

Poh-Sun Goh

28 June 2023, 1120pm, Singapore Time


As we experiment with, and develop a feel of the role, and value add of Artificial Intelligence (AI), engaging in a broader discussion of how we augment human intelligence can be helpful.

Why do we learn? How do we learn? What do we do when we have a question? Or need to find out how to do something? At the workplace, or in social and recreational settings? Empirical observations, reflection on personal practice, and focused interviews at the workplace bring up common actions - i.e. which are to look it up online or in a guide, ask someone (e.g. a peer, more experience colleague, or expert), read around the topic and do background research, or get training (e.g. short course, or longer training program).

When we look something up online, background AI algorithms including autocomplete text options in the search box, and sections of common similar and related questions and answers, as well as online resources, articles and reviews are served up to us on the first page of answers, reflecting what has been found useful for others (by viewership, engagement, time spent, click throughs and recommendations/citations). As we engage further with online content, background AI algorithms designed to keep us online will continue to refine what the search engine and online pages present and serve up to us. In this way, we are already taking advantage of AI, either consciously or unconsciously, on a regular basis, whenever we look something up online on a search engine, and progressively engage with content. This process mimics what happens when we 'ask someone', a peer, or experience colleague. In the case of online searches the process is massively scaled up taking advantage of a large pool of previous and ongoing topic searches, progressively refined by our further search and engagement behaviour. 

Embedded co-pilot AI within familiar professional and productivity software, and mobile Apps can potentially increase our productivity, by offering us first cut templates, initial drafts, and suggestions, drawn from what these AI models and tools have been trained on. As human users we then add individual and specific task based customisation and personalisation. Where and how AI adds value, and understanding how, when and to what extent we use AI will be an ongoing, iterative process. As we seek to get the job done, as efficiently and effectively as possible. Ultimately AI is simply another tool in our toolbox or toolkit. To be used, or not. Depending on what we need to do, and what the situation demands. 


Further Reading

Christensen CM, Hall T, Dillon K, Duncan DS. Know your customers’ “jobs to be done”. https://hbr.org/2016/09/know-your-customers-jobs-to-be-done. Accessed 28 June, 2023.

Gartner Hype Cycle. https://www.gartner.com/en/research/methodologies/gartner-hype-cycle. Accessed on 28 June 2023.

Goh, PS. Medical Educator Roles of the Future. (2020). Med.Sci.Educ. 30 (Suppl 1), 5–7. https://doi.org/10.1007/s40670-020-01086-w

Goh, PS. (2021). 'The vision of transformation in medical education after the COVID-19 pandemic'. Korean J Med Educ. 33 (3): 171-174. https://doi.org/10.3946/kjme.2021.197

Tuesday 27 June 2023

Integrity Concerns of AI in medical education scholarship

Integrity Concerns of AI in medical education scholarship

Mildred López


The introduction of artificial intelligence (AI) in learning settings has been a recent debate in educational institutions. The discussion used to be if institutions should allow these technologies to be part of campus life, just as we discussed if social media should be part of academic life. The different social media platforms not only became part of it but are where many interactions of the educational community take place. As institutions from different educational levels banned ChatGPT use in their learning settings because of concerns with academic integrity (Johnson, 2023), the question we should be asking is not whether we should allow it but how we will lay the ground rules for its use. 

Some of the arguments behind the controversy of the use of AI are ethical concerns, for example, how can we prevent students from plagiarizing using AI (Cotton et al., 2023). The general concerns of its use are described by Busch et al. (2023) with an analysis using the principles of biomedical ethics describing the need for autonomy for users to have transparency of the algorithms in models, justice on the equitable access of applications by all users, non-maleficence in the use of AI with critical thinking, and beneficence through the training of users before engaging with AI. Masters (2023) delves into the concerns from the development part of AI models; for example, it might be impossible to assure anonymity and privacy in the use of data that users provide through interactions with Chatbots. Another example described by the author is incorporating the algorithms with Learning Management Systems (LMS) that could provide feedback and personalized recommendations for learners, but how to ensure the data protection standards.

But what if the use of AI is not by the students but rather by academics? The concern migrates from the ethical dimension to more of the integrity behind scholarly engagement. Cotton et al. (2023) present an example of a paper where ChatGPT was used to lay ideas and draft the article but pose the question if it can be considered an author if it cannot take responsibility and not hold accountable. Hosseini et al. (2023) present a policy draft that focuses on disclosing the use of AI in tasks such as getting ideas or writing full text and accepting full responsibility for the submitted text. The discussion is only starting, but few developments are setting precedents for what academia should embrace to address the integrity concerns of AI in medical education scholarship.


References

Busch, F., Adams, L., Bressem, K. (2023). Biomedical Ethical Aspects Towards the Implementation of Artificial Intelligence in Medical Education. Medical Science Educator. https://doi.org/10.1007/s40670-023-01815-x

Cotton, D., Cotton, P., Shipway, R. (2023). Chatting and cheating: Ensuring academic integrity in the era of ChatGPT, Innovations in Education and Teaching International, DOI: 10.1080/14703297.2023.2190148

Hosseini, M., Rasmussen, L., Resnik, D. (2023). Using AI to write scholarly publications, Accountability in Research, DOI: 10.1080/08989621.2023.2168535

Johnson, A. (2023). ChatGPT In Schools: Here’s Where It’s Banned—And How It Could Potentially Help Students. Forbes. https://www.forbes.com/sites/ariannajohnson/2023/01/18/chatgpt-in-schools-heres-where-its-banned-and-how-it-could-potentially-help-students/?sh=1aea888a6e2c

Masters, K. (2023). Ethical use of Artificial Intelligence in Health Professions Education: AMEE Guide No. 158, Medical Teacher, 45:6, 574-584, DOI: 10.1080/0142159X.2023.2186203




Monday 26 June 2023

Role of AI in Health Professions Education

Role of AI in Health Professions Education

Poh-Sun Goh

27 June 2023, 0345am, Singapore Time


Artificial Intelligence (AI) is not only a "hot" subject, AI in Health Professions Education (HPE) is topical, trending and trendy. There is a sense that AI has reached a tipping point for widespread adoption, through open access websites (https://chatgpt-online.ai/), mobile Apps (https://openai.com/blog/introducing-the-chatgpt-app-for-ios), and increasingly being embedded as co-pilots in commercial and familiar enterprise software and platforms (https://sway.office.com/BnO7YrlOdmcDNp8c?ref=Link - see AI in Education section).

Fundamentally, AI can 'augment' human intelligence, through intelligent searches, e.g. AI algorithms underpinning online search engines guiding inquiry toward what other users have found useful and relevant (through number of views, time spent engaging in content, click throughs, recommending this content to other interested users and 'citations' or linking to search results); and being 'an instructor or tutor at our side' (Goh, 2020). Increasingly accessible, usable and workable AI will complement and selectively supplement our human role as instructors, educational guides, trainers, mentors and coaches; though it is likely that we will retain a distinctly human role if we focus on deeper individual customised engagement, to facilitate deep learning, by investing time, energy and effort to understand our human learners. This will require us as educators to skill up, and build not only digital literacy in AI, but also distinctly human skills and insights, into people (Shorey, Ang, Chua and Goh, 2022). 

It is very likely that the speed of individual and collective active exploration into the use of AI, building up empirical experience, use cases and examples, developing principles of professional practice and guidelines on proper usage, building theory, engaging in iterative rapid innovation using applied design thinking - ultimately refining the principles, practice and pedagogy of AI, will be faster that the cycle times for traditional peer reviewed publications. There will be a major role for the professional use of social media (including blogging) as content creation, curation and aggregation platforms to engage members of a professional Network, and Community of Practice (CoP). The dynamic and potentially more responsive nature of newer forms of digital scholarship and Micro-Scholarship (Goh, Roberts-Lieb and Sandars, 2022) will complement and add to traditional scholarship venues of academic and professional conferences, and peer reviewed publications, as we build up our experience and confidence in the use of AI in HPE, and understand its role in HPE.


References and Further Reading

Goh, PS. (2020). Medical Educator Roles of the Future. Med.Sci.Educ. 30 (Suppl 1), 5–7. https://doi.org/10.1007/s40670-020-01086-w

Sandars J., Goh PS. (2020) 'How to make it work: a framework for rapid research to inform evidence-based decision –making about the implementation of online learning during the COVID-19 pandemic', MedEdPublish, 9, [1], 154, https://doi.org/10.15694/mep.2020.000154.1

Goh, PS. (2021). 'The vision of transformation in medical education after the COVID-19 pandemic'. Korean J Med Educ. 33 (3): 171-174. https://doi.org/10.3946/kjme.2021.197

Shorey, S., Ang, E., Chua, J., & Goh, P. S. (2022). Coaching interventions among healthcare students in tertiary education to improve mental well-being: A mixed studies review. Nurse education today, 109, 105222. https://doi.org/10.1016/j.nedt.2021.105222

Goh, P. S., Roberts-Lieb, S., & Sandars, J. (2023). Micro-Scholarship: An innovative approach for the first steps for Scholarship in Health Professions Education. Medical teacher, 45:3, 307-312. https://doi.org/10.1080/0142159X.2022.2133689