The USA Junior AI Olympiad Is Here: What K–8 Families Should Know for 2027

In 2027, USAAIO runs its first Junior AI Olympiad, for students in grade 8 or below. Here is who can enter, what it tests, how the exam works, and how we think about preparing. Details are from the official USAAIO site as of November 2026, so check there for updates.

What it is

The USA Junior AI Olympiad (USA-J-AIO) is a new AI competition from the USA AI Olympiad (USAAIO), the organization that runs the high school USA–North America AI Olympiad. 2027 is the first year it will be held.

It's an individual competition with a single round. All 40 problems are multiple choice, and students don't write any code.

Key facts for 2027

- Exam date: Friday, May 7, 2027, 12:00 to 1:15 p.m. Eastern Time

- Registration deadline: April 30, 2027, 11:59 p.m. Eastern Time

- Eligibility: full-time students in grade 8 or below who are under 16 years old by August 30, 2027. High school students can't enter.

- Format: 60 minutes, 40 multiple-choice problems, maximum 300 points

- Awards: Distinguished Honor Roll, High Honor Roll, and Honor Roll

What it tests

The syllabus is broader than many parents expect for this age group. It covers four areas:

- Math foundations for AI: basic linear algebra, such as matrix multiplication and vector inner products (dot products), and basic probability and statistics, such as mean and variance

- Basic coding: reading and tracing simple Python, and explaining the logic of pseudocode. Writing code is not required.

- Machine learning: supervised learning (linear and logistic regression, support vector machines, decision trees, k-nearest neighbors, ensembles, bias and variance, cross-validation, and loss functions) and unsupervised learning (k-means clustering and principal component analysis)

- Deep learning: multilayer perceptrons, transformers, large language models, computer vision, and generative AI

That list looks intimidating for a ten-year-old. But the way the exam asks about it matters more than the list.

What it really rewards: understanding, not memorization

USAAIO says most problems are designed to be solved by reasoning from first principles. Students need to understand the core idea behind a method, why it works, and what its output means.

The official example is a good one. A student might compute a dot product, which is simple arithmetic, but should also understand *why* a dot product can measure how similar two things are, and how an image or a sentence can be turned into a list of numbers.

The exam doesn't require memorizing library commands or reading NumPy or PyTorch code. That's a deliberate choice. A student who has only used AI tools as a black box will find it hard. A student who understands what's happening inside them has a real chance.

How the exam works, and why parents should plan ahead

The logistics are serious, and they're the part families are most likely to overlook.

- It must be proctored onsite. Families arrange their own proctor, through a school, university, library, or testing center.

- Everyone takes it at the same time, wherever they are in the world. Noon Eastern Time is early morning on the U.S. West Coast and late evening in parts of Asia.

- The computer setup has rules. Students must use the contest app with lockdown enabled, run the required monitoring software, record their screen and face with OBS Studio, and use one monitor only. Leaving the contest app can lock the exam.

- There are no breaks and no reference materials. Scratch paper comes from the proctor.

- There's a follow-up deadline.Students submit links to their recording and other documents by 11:59 p.m. Eastern Time the next day.

- Unanswered questions earn partial credit. A correct answer earns 7.5 points, a wrong answer earns 0, and a blank earns 1.5. Students should understand this before exam day.

For younger students especially, a full practice run with the real setup is worth doing well before May.

Why it's worth a look

It's an early start on real AI understanding. As we found in our post on AP Calculus and the High School AI Olympiad, much of what the High School AI Olympiad tests isn't taught in a standard school math track. Students who build intuition for vectors, data, and models in middle school arrive in high school with a head start.

It suits younger students. Multiple-choice and no coding mean a student doesn't need years of programming to take part. What they need is curiosity and a willingness to think carefully.

It's a goal to learn toward. A fixed date gives a student a reason to go deeper than a casual interest would.

Our impression is that the best preparation isn't drilling practice questions. It's learning the ideas properly: what a model learns from data, how it can be wrong, and why. That understanding lasts long after the exam.

Things to note

This is the first year of the competition, so there are only sample problems, no past exams, and no record of how scoring will work out in practice.

If your student is curious about AI

We've taught AI to K–12 students since 2019. If you'd like to talk about whether the Junior AI Olympiad is a good fit for your student and how to get ready, we'd be glad to help.

Book a free consultation here.