The case for STEM in early education as AI rapidly changes tech landscape

Today’s generation of software-building, stack-maintaining IT pros regularly leverage their extensive expertise and problem-solving abilities.

You don’t acquire that kind of knowledge overnight. Christine Cunningham, chief curriculum and learning officer at Boston’s Museum of Science, wants to start building young engineers’ analytical acumen as early as preschool—and with materials as high-tech as puppets and pom-poms.

Cunningham leads the creation of instructional resources for a program called Youth Engineering Solutions (YES). Established in 2024, the initiative provides educator support and science, technology, engineering, and math (STEM) education for students in middle-school grades and below.

Preschoolers following the YES lessons are tasked with helping a puppet raccoon celebrate a birthday in style: given a jewelry box on a popsicle stick, the youngsters see which combination of paper clips, pom poms, sponge pieces, and craft foam make for maximum sound effects.

The game offers students a chance to experience the core steps of an engineering task: explore, create, and improve.

“Designing with a function in mind is something that’s not always present in preschool, but kids love doing it, and they’re very, very capable,” Cunningham told us.

With YES, Cunningham aims to teach students “durable” skills that advance with age: How to ask questions, consider a problem’s constraints, test ideas, evaluate evidence, and become persistent, adaptable, and confident in their approach to problem solving.

Evaluating results and working through problems are especially important as technologies like AI have shaken everyday infrastructures and workflows.

“Those kinds of experiences we think cultivate the habits of mind that will help students thrive in whatever the future holds for them,” Cunningham said.

YES please! Youth Engineering Solutions stemmed from a 2003 program, also led by Cunningham, known as Engineering is Elementary (EiE). YES expands beyond elementary age groups.

Cunningham, a former professor in education at both Tufts University and Penn State University, is also a curriculum specialist on Congress’s National Assessment Governing Board—a nonpartisan group tasked with measuring US students’ knowledge in math, reading, and other report-card subjects.

Machine learning by doing. A relatively recent addition to the curriculum calls for middle schoolers to train a computer to analyze fictional user reviews of slippers. With MathWorks’s MATLAB data-analysis software, the young students investigate how adjusting the training data—and assigning keywords with positive and negative attributions—impacts the machine learning model’s accuracy. Kids try again and again, fine-tuning the machine-learning model.

“Kids start to understand that AI is a tool made by humans,” Cunningham said.

YES’s approach to curriculum, according to Cunningham, does not emphasize how to use a given LLM or even how to code from scratch. What matters is how a student processes the results or how a student interprets the code.

“We give some students some basics around actually producing code, but more importantly, we’re helping kids think about, ‘What is computational thinking?’” Cunningham told us. “How do you define a problem correctly, and what are the criteria that make a project successful? When do you question the output? How do you debug something? How do you think about potential biases?”

The future is uncertain. A recent Pew Research Center survey of over 3,400 US adults found that more than half (52%) of US adults are more concerned than excited about the increased use of AI in daily life, particularly when it comes to issues like job loss. The workplace is likewise evolving its tech stack; a Gallup poll revealed that 47% of US employees stated that their employers have integrated AI tools, an increase from 41% last quarter.

With AI’s growth, how will starting STEM education early help prepare students, especially when the world and the IT industry might look radically different in a short period of time? The World Economic Forum, which polled employers for its 2025 Future of Jobs Report, found that top sought-after skills were:

• Analytical thinking
• Resilience, flexibility, and agility
• Leadership and social influence
• Creative thinking
• Motivation and self-awareness

Those soft skills and analytical abilities—which surpassed specific skills like cybersecurity, programming, and marketing, among the surveyed employers—feel similar to the ones being taught to the young noisemakers of the world.

Cunningham said AI can generate data and automate tasks, but “what it cannot do is replace human curiosity, or even the ability to define what is a meaningful problem that’s worth solving.”

So, Cunningham’s group now works on practicing those capabilities from an early age and creating those durable—and seemingly in-demand—competencies: “Developing that will give them a set of skills and that mindset to attack whatever might come in their future.”