Tiếp theoHướng dẫn tiếp theo
Dự án AI Scratch cho Trẻ em
Ứng dụng
HƯỚNG DẪN ứng dụng
A strong AI science-fair project asks a testable question about a system, its data or its effects, then compares results with a clear baseline.
Nghe hướng dẫn nàyAI thuật lại · ~3 min
Students should choose a safe scope, document their own work and check the current fair rules before collecting data or using AI in the submission.
An AI science-fair project should investigate a question, not merely demonstrate a tool. A useful question identifies what will be compared or measured: Does a classifier make more errors when the background changes? Does a small training set perform differently from a larger one? How does a simple rule-based baseline compare with a learned model on the same test examples? Keep the scope small enough to repeat and explain. Define the task, dataset, labels, baseline and evaluation method before running experiments. Separate training examples from test examples, record model and tool versions, and preserve a log of changes. Report failures as well as successes. If the project uses a public dataset, read its documentation and license; “available online” does not automatically mean unrestricted. Avoid using private or identifiable data without appropriate approval. Projects involving surveys, interviews, testing by other people or identifiable information may trigger human-participant review. The Society for Science’s ISEF rules explain that some student studies need prior Institutional Review Board review and consent, and that affiliate fairs can have additional requirements. Do not start data collection until the applicable teacher, sponsor or review committee confirms the plan. Avoid medical diagnosis, sensitive personal data or risky testing as a student project. AI use in the competition submission also has rules. Current ISEF guidance says AI may be used as a project resource with citation and acknowledgment, but prohibits generative AI from writing the research plan, abstract, poster or citations. Other fairs may set different rules. Check the current official organizer rules before beginning, keep the student’s own research and writing visible, and cite data, code, tools and collaborators. A successful project explains what the model did, where it failed and what the evidence can—and cannot—support.
Compare a simple rule-based baseline with a classifier on a small, public, permitted dataset and report the errors each makes.
Test how changing background or lighting affects a model’s image labels using non-identifying objects rather than people.
Measure whether a speech recognizer transcribes a set of student-written sentences differently in quiet and noisy conditions, with appropriate permissions.
Study a public dataset’s documentation and evaluate whether its labels and collection context fit a proposed use.
Science fairs are updating guidance as AI tools change, so students and sponsors should consult the organizer’s current rules early in project planning. Future projects may examine multimodal models, environmental costs or how people interact with AI, but strong research will still need a narrow question, safe data practices and transparent methods. Generative tools can support coding or exploration where permitted, while the student remains responsible for research decisions and presentation. An honest account of limits and mistakes is stronger science than an unsupported claim that a model works for everyone.
Bản tin miễn phí
Ba câu chuyện AI đã được xác minh vào mỗi buổi sáng các ngày trong tuần, được viết bằng tiếng Anh đơn giản. Miễn phí mãi mãi, không có quảng cáo.
Một email mỗi ngày trong tuần. Hủy đăng ký trong một cú nhấp chuột. Chúng tôi không bao giờ bán hoặc chia sẻ địa chỉ của bạn.
Tự kiểm tra
Phản hồi tức thì về mọi câu trả lời và chứng chỉ có thể chia sẻ với ID có thể xác minh sau khi bạn vượt qua khóa học.
Hỗ trợ giáo dục AI miễn phí. AI Understanding là tổ chức phi lợi nhuận 501(c)(3) — không bao giờ có quảng cáo, không có tường phí. Hãy quyên góp
A strong AI science-fair project asks a testable question about a system, its data or its effects, then compares results with a clear baseline. Students should choose a safe scope, document their own work and check the current fair rules before collecting data or using AI in the submission.
A bounded comparison identifies a condition and measurable outcome.
Held-out examples provide a more meaningful check of generalization.
Some human-participant research requires prior review and consent; the plan should be checked first.
Public availability does not itself establish permission or suitability.
A baseline helps interpret whether a model adds performance beyond a simple method.
Tiếp tục học hỏi
Đã chọn thêm hướng dẫn cho chủ đề này
Tiếp theoHướng dẫn tiếp theo
Dự án AI Scratch cho Trẻ em
Ứng dụng