JUDGING
Open registration, no cap. 3–5 finalist teams per track in each division present live. Every valid submission receives a certificate.
| Criterion | Weight | What judges ask | What a 3 looks like | What a 5 looks like |
|---|---|---|---|---|
| Creativity | 30% | Does the idea show your own observation, choices or a new way of expressing something? | A clear idea with at least one choice that is truly yours, not just the first thing the AI gave you. | A fresh angle judges haven't seen, and you can show where it came from. |
| Understanding & independent work | 25% | Can you explain your work, what the AI does and the key decisions you made? | Every member can explain what they made, what the AI does and one key decision. | You can explain why the AI behaves the way it does, including its mistakes, and the log clearly shows the students made the decisions. |
| Completeness | 20% | Does it meet your division's requirements, can you demo it, and did you test it? | Every Minimum to qualify item is done, and the demo works. (If any item is missing, this can't go above 2.) | All of that, plus testing beyond the minimum, with clear, checkable results. |
| Communication & teamwork | 20% | Is it clearly explained, and is the work sensibly shared? | Intro and talk cover the idea, the demo, the AI's job, the tests and the limits. Solo teams: well organised. | Clear and engaging for a non-expert. Every member contributes. Answers are specific. |
| Responsible AI use | 5% | Are sources and help clearly disclosed? Did you respect privacy and the limits of the tools? | Every tool, source and helper is credited, and the privacy rules are followed. | You spotted a real risk (bias, privacy, wrong answers), did something about it and can explain what. |
Each criterion is scored 0–5: 3 means your project meets your division’s requirements, 4 is clear and thorough, 5 has a clear highlight judges can check, and 0–2 means parts are missing. Each criterion’s score = raw score ÷ 5 × weight, so the total is out of 100. Example: raw scores 4, 4, 3, 4, 3 become 24 + 20 + 12 + 16 + 3 = 75. Lines of code, device cost, number of tools, paid plans, project size, hours spent, polished English, accent and language choice don’t earn or lose marks.
The criteria are the same in every division; each division is judged separately, with its own finalists and awards. A team that plays up is judged against that division’s requirements, with no bonus or penalty for being younger.
We first check eligibility and format, then at least two reviewers score every valid submission independently on the five criteria; Understanding and Communication & teamwork are judged from your intro and process log. The Boss Check is where reviewers start reading; it isn't extra credit. If two reviewers' totals differ by 20 points or more, a third reviewer scores it too, and we average all the reviewers' scores. The optional Reach Bonus (below) is added after that, by a separate team; reviewers never see it. Screening scores and rankings aren't published or shared with families.
In each division, 3–5 finalist teams per track are chosen, and the academic committee signs off the list. Every team hears by email at 09:00 SGT on Wed 13 Jan.
Three judges score each finalist team independently, starting fresh rather than carrying over screening scores, and we average their scores. If judges' scores differ a lot, we re-check the facts, but we never replace independent scores with a group discussion. Questions are friendly and focus on what you did and why.
Awards for every division are presented. Ties are decided by Understanding & independent work, then Creativity, then a review by an independent judge using the same criteria.
OPTIONAL
Building with AI includes getting people to see what you made. If you share your build publicly, your post can earn a small bonus in the online screening. It's optional: nobody has to post, and teams that don't post are judged exactly the same on the five criteria.
TOP 5 POSTS IN EACH DIVISION · SCREENING POINTS
Each division has three judges, from education, AI technology and creative fields. If a division needs a second judging zone, each zone keeps the same mix. At most one judge per division may come from a partner organisation; the rest are independent. Judges step aside from teams that include relatives, teams they have coached or taught this school year, or any other direct interest. Bilingual judges support presentations in Chinese. Judges are listed here from Fri 18 Dec, and added as they confirm. Screening reviewers come from our academic committee, our judges, and vetted teachers and industry reviewers; they follow the same code of conduct and step aside from teams they know. Finals judges don’t screen entries from the division they judge live, and a partner representative never reviews an entry alone.
| Finalist teams in the division | Places |
|---|---|
| 5 or more | Champion, 1st Runner-up and 2nd Runner-up |
| 3–4 | Champion only |
| 1–2 | Teams present, without ranked places |
Prizes: Trophies and cash prizes; amounts announced once funded.
Awards are given to finalist teams within each division across all four tracks, up to nine places in total. There are no track awards and no overall champion. Awards may be left unfilled if there’s a serious breach of the rules or no project meets the minimum requirements; we record the reason. Prize money is paid per team within 30 days of the event (by Tue 16 Feb): each winning team names one person to receive it (a parent receives it for a minor), and the team agrees how to share it. We collect payment details only from winning teams, after the event.
Community Choice is voted in the room on Showcase & Awards Day (Sun 17 Jan) by checked-in participants, one vote per project: one award per division, for a team’s project, not a person. YAI Ambassador recognition thanks students aged 13+ who shared YAI with friends. Neither affects judging, finalist selection or the judged awards, and neither includes cash. Community Choice rules are on the Demo Weekend page; YAI Ambassador rules are published by Fri 30 Oct.
Every valid submission receives a certificate: each team member gets a participation e-certificate on Fri 22 Jan, whether or not they come to Demo Weekend. Finalists also receive a finalist certificate, and winners an award certificate; printed copies are handed out at Demo Weekend, and e-copies follow on Fri 22 Jan. Certificates show the display name you confirmed at registration. They record taking part, reaching the finals or winning an award. They are not an official qualification, and YAI doesn’t promise any school or university admission, scholarship, or place in another programme or competition.
We don’t provide individual or team feedback reports, and screening scores and rankings aren’t shared. All criteria are published here so you can check your own work, and for finalists, the judges’ questions on the day are part of the experience.
Appeals can cover eligibility, scoring errors, conflicts of interest or how a rule was applied. Disagreeing with a judge's opinion is not grounds for an appeal. Screening appeals: by Thu 14 Jan, 12:00 SGT, to support@ycess.com. Someone who didn't review your project checks it and replies the same day, by Fri 15 Jan, 12:00 SGT at the latest. If an appeal is upheld, the team is added as a finalist and no one else loses their place. Result appeals: within two working days after the awards (by Tue 19 Jan, 23:59 SGT). Someone not involved in the original decision reviews it, and we reply by Fri 22 Jan.
We’re looking for judges from education, AI technology and creative fields, and screening reviewers who are comfortable reading student projects online.
Everyone signs our code of conduct, declares any connection to a team and steps aside from teams they know. Judges never contact students privately.
Tell us about yourself: email partner@ycess.com with “Judging” in the subject line.