A novel artificial intelligence leadership curriculum for pre-college students interested in medicine and engineering: program development and global competition outcomes

Introduction

Artificial intelligence (AI) has revolutionized how we analyze and solve problems in many fields, including business, medicine, and engineering (1-4). One example of how AI has revolutionized is in therapeutic drug discovery (5). Drug discovery is a complex process and relies mostly on trial-and-error. AI has been used to optimize the process by enabling more efficient and accurate analysis of different potential drugs and their toxicity. Another powerful application of AI is in the field of agriculture. For example, the Agrobot E-Series robots use an onboard short-range integrated color and depth sensor to evaluate the ripeness of strawberries. It then uses its robotic arms to pluck and store the berries, increasing the efficiency of this agricultural process. Given AI’s continued and exponential impact in many fields, it is becoming increasingly more critical to familiarize students with this technology, including at a pre-college level (6). The understanding and familiarity of AI foundational concepts will allow pre-college students to apply these skills to their specific area of study. Unfortunately, the coronavirus disease 2019 (COVID-19) pandemic and lockdown undoubtedly had an impact on education worldwide, including a decrease in in-person opportunities for students to have hands-on learning for an optimal educational experience (7). Prior to the pandemic, there was a momentous emphasis on providing the future generation with interactive opportunities to further develop their interests and ambitions. These opportunities included laboratory research, in-person volunteering, team athletics, and internships that allowed for professional and personal growth (8,9). However, with the COVID-19 lockdown, many students experienced a reduction in interactive, hands-on learning experiences to augment personal growth such as leadership, interpersonal skills.

Nevertheless, with new challenges come novel solutions; we were very fortunate to host the first LeadingAI DeepRacer training program, a course designed to have students experience immersive learning with machine learning (ML) with the goal of building leadership skills. This training program brought together high coaches from three different states: Hawaii, California, and Ohio. Despite the seemingly different experiences and backgrounds of the individuals that made up the coaching team, we were unified under the mission of discovering how to apply our knowledge of AI to serve the students. The high school students underwent mu training sessions that primarily focused on two topics: AI/ML and leadership development.

In this report, we describe the curriculum development and 3-year outcomes following the initiation of this program. The curriculum is multi-faceted containing various components including hands-on learning, interdisciplinary didactics, leadership curriculum, team-based learning, and competition experiences (Figure 1). We report qualitative feedback from students about the curriculum including their thoughts on how this program can be merged into their fields of interest/future college majors. We also discuss the competition results. This program successfully trained students for the AWS competitions with multiple students placing in the top percentages of global rankings and championships. The purpose of this report is to serve as a curricular foundation for other educators to model an AI curriculum for pre-college students, as well as highlighting the ability of pre-college students to perform at some of the highest stages of competitive AI following this novel curriculum.

Figure 1 The multidisciplinary curricular components of the AI Leadership Course facilitated a holistic curriculum, enabling exploration of AI within the context of various career paths. AI, artificial intelligence.

Methods

LeadingAI leadership and curriculum

What is Amazon Web Services (AWS) (10)

AWS is the world’s most comprehensive and broadly adopted cloud, offering over 200 fully featured services from data centers globally. Millions of customers—including the fastest-growing startups, largest enterprises, and leading government agencies—are using AWS to lower costs, become more agile, and innovate faster.

What is AWS DeepRacer and the goals of the DeepRacer program

DeepRacer is a platform that allows developers to learn about and experiment with reinforcement learning (RL) through an autonomous race car. The car is a 1/18th scale model that is controlled by a computer and can be programmed to navigate a physical or online track using ML algorithms. AWS DeepRacer is designed to be easy to use for developers of all skill levels, with a focus on making it easy to learn about and experiment with RL. RL is a type of ML technique that enables an agent to learn in an interactive environment by trial and error using feedback from its own actions and experiences.

The coaching team

The coaches consisted of volunteers from multidisciplinary backgrounds including physician, software engineer, cloud architect, research scientist, consultant, medical student, and life coach. Senior coaches include individuals that held diverse roles within their own respective careers such as a solution architect and a degree in executive leadership. The diverse range of backgrounds enabled specialized expertise training in each component of the training program.

It was critical to provide training for incoming coaches due to the various levels of expertise in DeepRacer. Before the start of the training program, the senior coaches provided training sessions for all the coaches. These sessions covered the following topics:

• The vision and mission of the training program;
• The technical details of DeepRacer and how to race DeepRacer online;
• Understanding the students and how to motivate them;
• The StrengthsFinder and how it can help students discover their strengths.

For each training program, we divided the students into small teams, with each team having 3–5 students. Each small team had two adult coaches and 2–3 assistant college coaches. The coaches ensured that each student was progressing and not falling behind in their learning. During each weekly session, every student attended the material together in a large group setting. After the large group meeting, the students would divide into their small teams to discuss and ask questions about the issues they faced while training their DeepRacer models.

Recruitment of students

The initial phase of recruitment was primarily dissemination through word of mouth among the respective local communities. Several cities that had a higher concentration of students, a physical track was built for in-person engagement events which served as another form of recruitment to the local communities.

After the students apply for the program, the senior coaches will interview every applicant to make sure they have the aptitude and passion to learn DeepRacer and leadership. We found out that some students applied for the program because their parents wanted them to be there but they didn’t have a desire to join the program. Some students were already too busy with their academic work and extracurricular activities that they would miss a few learning sessions. For all those students, we would encourage them to apply in the future when they are ready.

We ran the online training program twice a year. From 2021 to 2023, we conducted a total of six training programs in these 3 years. The high school students came from California, Hawaii, Ohio, Florida, New York, North Carolina, Texas, Maryland, Nebraska, Tennessee, Pennsylvania, and Massachusetts, some also came from Canada and China. We had a total of nine students trained in the 3-year period.

Curriculum components

Our vision for this training program is to train young leaders to use AI for helping themselves and serving others. The integration of evidence-based instructional practices within curriculums has been shown to contribute to longitudinal and life-long learning (11). We have a three-prong strategy to help to achieve our vision (Figure 1):

• Establish AI skills and develop characters of teamwork and cooperation through learning AI and related tools;
• Built-in core values: authenticity, service, excellence;
• Project based learning, small group training, active learning.

For the AI skills, we have two sessions. One is “Artificial Intelligence: Past, Present, and Future” with the following topics:

• Past:
  • The birth of AI;
  • First AI winter;
  • Expert systems;
  • Second AI winter;
  • ImageNet.
• Present:
  • Deep learning/neural networks: overview;
  • Five challenges—deep learning.
• Future:
  • 4M (model, machine, mechanization, map);
  • Contrastive learning;
  • Hybrid AI;
  • Causality deep learning.

The other is “Generative AI” with the following topics:

• What is AI and generative AI;
• Interesting applications;
• Overwhelming risks;
• Recent advances;
• Generative AI as CoPilot;
• How to innovate with generative AI;
• Company & government policies.

The expected results of the training are:

• Students understand and apply RL in the real world;
• Train ML models and use the models for racing;
• Finish StrengthsFinder test and apply leadership skills in projects;
• Students who perform well will be eligible to become assistant coaches for the next DeepRacer training.

The AI Leadership Course featured multidisciplinary curricular components, including a diverse coaching staff from medicine, business, engineering, and social sciences. The course offered hands-on learning through the AWS DeepRacer competition, which involved in-person racing tracks, weekly coding and model training sessions, and participation in national and international competitions.

Additionally, the course included didactic sessions covering medicine, theoretical ML, and practical ML with DeepRacer. Leadership development was emphasized through activities such as StrengthsFinder analysis and public speaking opportunities. Many activities utilized a team-based learning approach to enhance teamwork skills in AI.

The primary goals were to understand AI foundations through in-person experiences, build confidence in applying AI tools within students’ fields of interest, and learn valuable teamwork and leadership skills in AI.

We began the DeepRacer training program in January 2021 and experimented with various curriculum durations. Initially, we conducted the training over 12 weeks, meeting for 2 hours each week. While we covered many different areas, feedback from students and coaches indicated that the training duration was too long. In response, we shortened the duration to 9 weeks, still meeting for 2 hours each week. Despite this change, some students continued to feel the training was too lengthy.

We then tried a more condensed format, running the training over 2 weeks with 6-hour weekly sessions. However, this shorter duration left students with insufficient time to practice running the DeepRacer. Consequently, we adjusted the program to 5 weeks, meeting for 3 hours each week. This 5-week schedule appears to strike a good balance, providing students enough time to engage thoroughly with the DeepRacer training (Figures 2-5). Overall, the 5-week curriculum resulted in the greatest efficacy and overall satisfaction for both the students and coaches. The students have ample time to learn the material and to have hands-on experience with the DeepRacer, but not too overburdened with time commitment.

Figure 2 Twelve-week training curriculum. Ample time to review all material and hands-on training. However, it is very time consuming for the students and the coaches. AI, artificial intelligence; AWS, Amazon Web Services.

Figure 3 Nine-week training curriculum. Despite condensing the 12-week curriculum into 9 weeks, we still received feedback suggesting the training should be shortened further. AI, artificial intelligence; AWS, Amazon Web Services.

Figure 4 Five-week training curriculum. The 5-week schedule is easier to manage since students need to commit for only a short period. Although still tight on time, it is sufficient for teaching the material and providing adequate training. This format strikes the right balance between time commitment and comprehensive learning. AI, artificial intelligence; AWS, Amazon Web Services.

Figure 5 Two-week training curriculum. This schedule was much easier to manage since coaches and students only needed to commit to 2 weeks. However, students struggled to concentrate for 6 hours per session, and we did not have enough time to cover the material and training adequately. AI, artificial intelligence; AWS, Amazon Web Services.

Results

StrengthFinder results

Each student and coach participate in the Gallup StrengthFinder survey to discover their strengths (Figure 6). After understanding their strengths, students can try to pursue opportunities in their future careers and education that optimize these strengths (12). By understanding each member’s strengths, the team can more effectively distribute the work according to the strengths of each member (Figure 7).

Figure 6 This is a sample reports of the strengths of the team members of each team. There are four areas of strengths—“Strategic Thinking”, “Influencing”, “Relationship Building”, and “Executing”.

Figure 7 In addition to understanding the strengths of individuals, it is important to understand the strengths of the entire team since each team member brings different strengths to the team.

DeepRacer RL curriculum

AWS DeepRacer is a fully autonomous 1/18th scale race car designed to test RL models by racing on a physical track. When setting up an AWS DeepRacer model, there are three critical concepts to understand: hyperparameters, action space, and reward function.

Hyperparameters

Hyperparameters define the behavior of the RL model but are not updated during training, influencing how effectively and quickly a model learns. In the context of AWS DeepRacer, key hyperparameters include:

• Learning rate: determines how much the model adjusts its predictions with each new piece of evidence. A higher learning rate might lead to faster learning but can overshoot the optimal solution.
• Exploration vs. exploitation trade-off parameters: these parameters manage the balance between exploring new actions to find better ones (exploration) and using actions known to yield high rewards (exploitation). This is typically managed through an epsilon-greedy strategy.
• Batch size: the number of experiences used in one iteration of model updates. A larger batch size can lead to more stable updates but requires more memory and computational power.
• Number of epochs: the number of times the learning algorithm will work through the entire training dataset.
• Discount factor: determines the importance of future rewards vs. immediate rewards.

Action space

The action space defines the set of all possible actions the agent can take at any given state. For AWS DeepRacer, the action space includes the different combinations of speed and steering angles that the car can adopt. A well-designed action space allows the car to navigate the track efficiently while avoiding actions that could lead to crashes or going off-track. Examples include:

• Different degrees of turning (e.g., slight left, hard left, slight right, hard right);
• Different speeds (e.g., slow, medium, fast).

Reward function

The reward function is arguably the most critical part of setting up an AWS DeepRacer model, providing feedback to the model about the desirability of its actions in given states. The design of the reward function directly influences the learning and performance of the model by guiding it toward the optimal policy for efficient track completion (Figure 8).

Figure 8 The reward function directly influences the desired behaviors (e.g., staying on the track, high speed on straightaways, efficient cornering) and penalizes bad behaviors (e.g., going off-track, collisions).

An effective reward function may include components such as:

• Rewarding the car for staying close to the centerline;
• Penalizing the car for steering too sharply at high speeds;
• Providing incremental rewards for progress made along the track.

In summary, hyperparameters tune the learning process, the action space defines what the car can do, and the reward function guides the car’s behavior towards the goal. Together, these elements form the backbone of a RL model for AWS DeepRacer, allowing it to learn from interactions with the environment and improve over time.

Competitions

Students were encouraged to compete in real-life national competitions to apply the knowledge and skills developed through the curriculum. The ability to perform and translate these practical skills to a competition not only demonstrates their technical abilities but also an understanding of how these skills can be applied to real life situations (13-15).

There are four levels of competitions:

Community race

A community race is a private event where any DeepRacer participants can create a model to compete. The community race is a very good format for novice DeepRacer participants to try out their models. The community race is also a way for the DeepRacer organizer for their local community.

Virtual Race

Any DeepRacer user can compete at the monthly Virtual Race at the national and regional level. The #1 racer of each of the six regions is eligible to earn the opportunity to participate in the AWS DeepRacer championships at AWS re:Invent.

AWS Summits

AWS Summits are in-person events that AWS organizes around the world to bring the cloud computing community together to connect, collaborate, and learn about AWS. One of the events in the AWS Summits is DeepRacer physical competition. Top racers will receive prizes and the fastest racer will win a trip to compete in the AWS DeepRacer Championship Cup at AWS re:Invent 2023.

AWS re:Invent

AWS re:Invent is an annual global event in Las Vegas. From March to October, the top 72 DeepRacer participants will emerge from AWS Summit and Virtual Race racing; earning a trip to AWS re:Invent in Las Vegas and a chance to compete for the Championship Cup.

Discussion

Lectures in medicine and AI

Many students enrolled in the program voiced an interest in medicine as a future career. It has been well documented in the literature that current and emerging AI techniques have revolutionized the field of medicine (16). These tools include various techniques including convolutional neural networks, generative adversarial networks, and large language models. As such, a goal of the program was to offer a unique opportunity for students to gain hands-on learning on technical AI skills while learning about current and potential applications in healthcare. Part of this integrative experience included a lecture and discussion regarding real-life applications of AI in healthcare with a physician. A physician involved with research in AI in medicine and LeadingAI coach gave lectures entitled “Fundamentals of Convolutional Neural Networks and Practical Applications in Medicine” and “Applications of Artificial Intelligence in Healthcare and Fundamentals in Convolutional Neural Networks”. These lectures included a clinically-relevant case study that emphasized how AI can be utilized to address certain longstanding barriers and gaps in healthcare. The lecture also included various applications of AI in medicine including automated detection of referrable disease, drug discovery optimization, and precision therapeutic development. The lecture concluded with a discussion with the students and coaches about their thoughts on AI in medicine. By providing real-time discussion between students and a physician involved in AI research in medicine, students interested in medicine were able to gain unique insights on how the fundamentals of their hands-on work with DeepRacer can be applied to future healthcare applications.

Qualitative feedback results and competition outcomes

In this section, we report the qualitative feedback from the students and assistant coaches regarding this program. We first start with the curriculum feedback from various students. As several of these pre-college students are entering college in the near future, several students discuss how the curriculum applies to their field of interest/major. It is of note that some students returned and led as coaches.

Qualitative curriculum feedback

Student 1 curriculum feedback

“LeadingAI’s AWS DeepRacer course is not a traditional computer science class. When I first saw the course’s overview, I was expecting three months of monotone lines of code and geeky discussions about python. I was completely wrong. By avoiding the traditional approach of brute-feeding students information, LeadingAI encourages students to critically think and be responsible for their own learning in the class. Because the coaches have inspired students to discover and troubleshoot by themselves, I have developed a genuine interest in AI and how it can be applied to various fields, like the medical field, which I aspire to go into.”

Student 2 curriculum feedback

“Throughout this DeepRacer class, I have gained a better understanding of AI and a better idea of what I want to do in the future. I had very little experience in Python from the start, but now I am able to understand the basics. One important lesson that I learned is that in order to create a satisfactory DeepRacer model, I have to keep trying and not give up. The StrengthsFinder test also helped me realize my strong points and gave me more confidence in pursuing a future career in the AI field. Broadening my knowledge, especially in this field, was a very worthwhile and exciting experience for me. I definitely look forward to participating in this class again.”

Student 3 curriculum feedback

“I liked this training because it is more hands-on than most school classes which helped me to gain a better understanding of the material that I was learning. I could experiment and use trial and error to figure out what worked which reinforces my own knowledge because I could see patterns and draw conclusions from those observations. Since I’m going into the Mechanical Engineering major and possibly field post-college, getting a good understanding of machine learning is important, especially understanding the concepts behind it. The leadership aspect was also useful in that it provided me with better knowledge of how I can apply myself to different situations and how I can be most effective. I think the training content and method went at a good pace, and I really enjoyed the smaller breakout sessions with our individual groups. We were able to communicate with each other and bond over similar interests in this training. This training is definitely important to do in high school because it lays the groundwork for students to become bigger leaders in college and stand out. They don’t need to spend time figuring out how they can best apply themselves in college if they learn that in high school. AI is the future of a lot of technology and understanding it will help give a headstart into that up-and-coming industry. Leadership is important to everyone because anyone can be a leader if they wanted to, and knowing what you are good at is a big step into becoming that great leader.”

Student 4 curriculum feedback

“When I first joined LeadingAI, I didn’t know how significant the Amazon Web Services (AWS) program was and would become in my life. We all know the exclusivity of traditional robotics clubs. The programmers may hog the programming part and your team captain may only permit you to work on certain tasks. On the contrary, LeadingAI is not limited by these roles. Coaches train you through small class sizes with like-minded individuals to teach you how to analyze, explore, and problem-solve. Unlike any other programming program out there, LeadingAI gives you the unique first-hand ability to engage and manipulate artificial intelligence online without any prior knowledge. In this fast-paced era and as a committed lifelong learner, I believe that learning how to harness artificial intelligence is an essential skill. Aspiring to become a mechanical engineer, being a participant in the AWS LeadingAI program has definitely opened my eyes to the possible applications of artificial intelligence in current and futuristic technologies. LeadingAI has taught me how to solve problems more efficiently and has brought me closer to understanding emergent technology.”

Student 5 curriculum feedback (later returned as assistant coach in a later season)

“I began my journey with the LeadingAI DeepRacer team as one of the first students to join the program. The unique nature of the challenges and opportunities offered by both the competition and the LeadingAI lessons allowed me to grow and greatly expand my abilities both as a programmer and a student and leader. The variety of opportunities to explore beyond the given tools to make customization and improvements to model training and development were an incredible experience in hands-on problem solving, from Linux installation to Selenium automation. As assistant coach, I’ve had the opportunity to work with many talented individuals and pass on what I have learned through the LeadingAI program and my explorations of AI. Getting to help others through the same challenges I once faced helps me reinforce the content and become a better teacher. Finally, I have even been lucky enough to race in person and apply the theoretical models to a real-world environment. The accompanying real-world challenges in hardware and software offer yet more chances to practice problem-solving skills and have a lot of fun in the process. I know that the abilities I have developed through LeadingAI will serve me excellently studying computer science at Stanford University. My love for STEM has been driven by hands-on projects such as Deepracer, and I’m excited to be able to experience more similar opportunities in the future.”

Assistant coach curriculum feedback

“I was in a unique position when first joining the LeadingAI DeepRacer course. I was an assistant coach but I was new to the concept of machine learning and the AWS DeepRacer. For me, it was interesting to think about how I can teach a robot. I am more used to laying out steps that the computer will execute rather than learn from. It reminded me what it takes to fully learn something. To learn means to be able to reteach the concept to a person (or in this case a computer) who doesn’t have prior background to the program. It has shown me that fully understanding something means to be able to break down what I know in simple terms in order for someone/something to understand. More than learning about machine learning however, being able to hear about the students and the coaches were all very interesting. Everyone has different backgrounds and seeing how they applied that to STEM helped open new horizons to possible careers in STEM. The teams were all supportive. It was a positive learning environment where everyone was happy to help one another get better. Finally, being able to say that I’ve had a background in reinforcement learning has allowed me to explore other opportunities. I’ll be joining a machine learning research lab over the summer of 2022 at the University of Southern California (USC) and I think that my experience with LeadingAI has definitely helped me.”

The student feedback aligns well with the expected outcomes of the training program. The key themes reflect the vision of training young leaders to use AI for personal development and serving others. The feedback indicates that the program successfully integrated evidence-based instructional practices, project-based learning, small group training, and active learning, fostering long-term and meaningful educational experiences.

Competition outcomes

2021

One student was a champion at a monthly US “Virtual Race” and was offered a full ride to compete at the 2021 AWS re:Invent (the largest event organized by AWS).

2022

One student and one coach were champions at two monthly US “Virtual Races”. They were offered full rides to compete at the 2022 AWS re:Invent.

2023

One student was a champion at a monthly US “Virtual Race” and was offered a full ride to compete at the 2023 AWS re:Invent. One coach was a champion at an AWS Summit in Europe and was offered a full ride to compete at the 2023 AWS re:Invent. One student was a champion at a monthly China “Virtual Race” and was offered a full ride to compete at the 2023 AWS re:Invent.

Challenges

Despite the success and accomplishment, the development of this novel program did not come without challenges. It was critical for the DeepRacer team to undergo continuous reflection to address these areas of improvements and challenges.

• Lack of motivation among students: some students lacked the motivation to learn the DeepRacer material and attended only because their parents wanted them to.
  • To address this, we implemented an interview process for the DeepRacer program. After applications were submitted, we arranged interviews for every applicant. Three senior coaches conducted these interviews, each responsible for one-third of the applicants. We rejected some applications after determining that the applicants lacked sufficient motivation to learn the material.
• Attendance issues: the training program lasted for 12 weeks, and some students couldn’t attend all the sessions due to other commitments.
  • We revised the program to offer 12-week, 5-week, and 2-week durations. We discovered that the 2-week program was too condensed and the 12-week program was too long. We concluded that the 5-week program was optimal in terms of duration and the amount of material we could teach.
• Training costs: students could train their DeepRacer on AWS with 10 hours of free training. After 10 hours, they had to pay for additional training.
  • To mitigate this, the coaches developed a “Local Server Installation Guide” to help students install the DeepRacer training software on desktops with Nvidia graphics processing units (GPUs). Additionally, some coaches installed DeepRacer models on their desktops, allowing students to send their models to these coaches for training.
• Financial constraints: some students might not be able to afford the tuition for the training program.
  • We implemented a tuition assistance program for applicants. If an applicant met the income criteria, we waived the tuition fee.

Conclusion and future directions

Throughout the DeepRacer training program, the high school students worked with AWS’s platform to train an Amazon DeepRacer through ML on AWS DeepRacer’s virtual racetrack. Throughout this training, students collaborated with each other and with coaches from multidisciplinary backgrounds including physician, software engineer, cloud architect, research scientist, consultant, medical student, and life coach to optimize their learning and their racing. In addition to the DeepRacer, lectures were given by coaches to give insight into the unique applications that AI can have in real life scenarios. These lectures were complemented with a research presentation by students on applying AI in different fields such as healthcare, business, cuisine, and gaming. The students were also given the opportunity to showcase the knowledge and skills gained by submitting their individual models to Amazon’s Global League competition and participate in inter-organizational competition in teams. These competitions are typically participated by working professionals, thus allowing these high school students to perform at some of the highest levels. The goal of the inter-organizational competitions not only recognized the progress and skills of individuals but also encouraged partnership among peers to optimize their results.

While there was a strong emphasis on the technical aspect of ML, one of the primary goals of this program was to help students identify their personal strengths. Each individual, including the coaches, took a StrengthsFinder test to gain better insight to their own unique leadership qualities. As the training program progressed, the coaches encouraged students to introspectively analyze their strengths and employ them during their group projects. By the end of the course, students were more confident in their abilities to navigate new challenges, such as utilizing AI for racing competitions. While some students expressed interest in careers outside engineering, it was largely agreed that these skills to work with others and learn new technology will be beneficial regardless of career field. The field of AI is rapidly expanding, and its utility as a diverse educational tool is highly effective. Future research regarding understanding this program further should be focused on quantitative performance metrics. This can be achieved through pre- and post- course evaluations, as well as DeepRacer performance at the beginning and end of the course.

Looking ahead, the LeadingAI programs aims to continue its mission of expanding its reach by equipping pre-college students with experiences and training to become leaders in AI in their respective careers. Efforts are being made to broaden this reach to international students, promoting access to AI education globally. LeadingAI hopes to position students to pursue their interest with the confidence to thrive and make a meaningful impact in this rapidly advancing technological field.

Acknowledgments

Funding: None.

Footnote

Peer Review File: Available at https://jmai.amegroups.com/article/view/10.21037/jmai-24-156/prf

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://jmai.amegroups.com/article/view/10.21037/jmai-24-156/coif). J.O. served as unpaid Associate Editor-in-Chief for Journal of Medical Artificial Intelligence at time of submission. D.C. is employed by Oracle. S.L. is employed by Google. D.O. is employed by Amazon. The other authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Xu W, Ouyang F. The application of AI technologies in STEM education: a systematic review from 2011 to 2021. Int J STEM Educ 2022;9:59. [Crossref]
2. Lyu Z, Ali S, Breazeal C. Introducing Variational Autoencoders to High School Students. Proceedings of the AAAI Conference on Artificial Intelligence 2022;36:12801-9. [Crossref]
3. Lee Y, Douglass A, Zeng S, et al. Preparing early educators as frontline leaders and change agents with a leadership development initiative. International Journal of Child Care and Education Policy 2022;16:2. [Crossref]
4. Abdelhalim H, Berber A, Lodi M, et al. Artificial Intelligence, Healthcare, Clinical Genomics, and Pharmacogenomics Approaches in Precision Medicine. Front Genet 2022;13:929736. [Crossref] [PubMed]
5. Blanco-González A, Cabezón A, Seco-González A, et al. The Role of AI in Drug Discovery: Challenges, Opportunities, and Strategies. Pharmaceuticals (Basel) 2023;16:891. [Crossref] [PubMed]
6. Chaudhry MA, Kazim E. Artificial Intelligence in Education (AIEd): a high-level academic and industry note 2021. AI Ethics 2022;2:157-65. [Crossref] [PubMed]
7. Catena RD, Carbonneau KJ. Guided Hands-On Activities Can Improve Student Learning in a Lecture-Based Qualitative Biomechanics Course. Anat Sci Educ 2019;12:485-93. [Crossref] [PubMed]
8. Al-Natour A, AlNatour A, Ali RA, et al. Students' perceptions and experiences in a health promotion course using interactive learning. Heliyon 2021;7:e07192. [Crossref] [PubMed]
9. Muncan B, Majumder N, Tudose N. From high school to hospital: how early exposure to healthcare affects adolescent career ideas. Int J Med Educ 2016;7:370-1. [Crossref] [PubMed]
10. Amazon. What is AWS. Available online: https://aws.amazon.com/what-is-aws/ [accessed 7/20//2024].
11. Elrod S, Kezar A. Increasing Student Success in STEM: Summary of A Guide to Systemic Institutional Change. Change The Magazine of Higher Learning 2017;49:26-34. [Crossref]
12. Murray WB, Henry J, Jackson L, et al. Leadership Training: A New Application of Crisis Resource Management and Distance Education in a Large Group Format at a Medical Simulation Facility. J Educ Perioper Med 2002;4:E021. [PubMed]
13. Karakose T, Tulubas T. School Leadership and Management in the Age of Artificial Intelligence (AI): Recent Developments and Future Prospects. Educational Process International Journal 2024;13:7-14. [Crossref]
14. Anastasiadis T, Lampropoulos G, Siakas K. Digital Game-based Learning and Serious Games in Education. International Journal of Advances in Scientific Research and Engineering 2018;4:139-44. [Crossref]
15. Wang LH, Chen B, Hwang GJ, et al. Effects of digital game-based STEM education on students’ learning achievement: a meta-analysis. International Journal of STEM Education 2022;26:9. [Crossref]
16. Rajpurkar P, Chen E, Banerjee O, et al. AI in health and medicine. Nat Med 2022;28:31-8. [Crossref] [PubMed]