Last updated: October 2026
AI writes the language and the ideas. Code produces and double-checks the answers. Sources supply facts when we use them. Teachers review.
Last updated: October 2026
Every part of an activity carries a level. Here is what each level means.
| Level | Label | Example |
|---|---|---|
| 1 | Computed by code | A bar model is drawn by code from the problem structure. |
| 2 | Computed twice and compared | An addition fact is computed, then checked by subtraction. Both agree. |
| 3 | From a cited source | A data value comes from a named source with a record id. (Reserved for future tools.) |
| 4 | AI-written, independently checked | A story sentence is AI-written, then an independent AI pass checks it for grammar and grade-appropriateness. |
| 5 | AI-written, review recommended | A prompt is AI-written and not independently checked, or an item was edited after checking. |
Answers checked by code
Every answer, data value, and drawing was computed by code and double-checked. All checks passed. Wording may still need your judgment.
Checked, teacher review recommended
Answers are checked, but some wording or pictures are AI-written and independently checked. Please review before using with students.
Needs review
A check failed, an item was edited after checking, or a check was accepted despite a problem. Please review before using with students.
7 + 8 = 15
The answer 15 was computed by code, then checked by subtraction: 15 - 7 = 8. Both agree.
Show the calculation
7 + 8 = 15
Strategy: make ten — 7 + 3 = 10, then + 5 = 15
✓ Subtraction check agrees: 15 - 7 = 8
A pencil is measured with a ruler.
Show the calculation
Each small mark is ¼ inch.
The object ends 25 marks from zero.
25 × ¼ = 6 ¼ inches
✓ Second method agrees
Maya had 12 apples. She gave 5 to Liam. How many does she have now?
Show the calculation
Structure: separate, result unknown
12 - 5 = 7 apples
✓ A second AI solved this without seeing the answer and got 7.
Story language is AI-written and independently checked.
Original: "the dog ran fast"
Show the calculation
Rule: Capitalize the first word
Corrected: "The dog ran fast"
✓ Round-trip test confirms the correction reproduces the clean text.
On any activity, tap the trust chip and then "Report a problem." Choose what is wrong, optionally note which page or problem, and add a short comment if you like. No name or email is stored. We review every report.
| Area | Level | What is checked |
|---|---|---|
| Words | Computed twice and compared | Every word is checked against the decodable word bank, target words, and allowed lists. A second independent check verifies decodability. |
| Wording | AI-written, independently checked | Sentence grammar and naturalness are verified by an independent AI pass. |
| Pictures | AI-written, independently checked | Picture keys come from the code library. A second AI pass verifies the picture matches the sentence. |
| Area | Level | What is checked |
|---|---|---|
| Answers | Computed twice and compared | Every answer is computed by code, then re-computed with a second independent method. Both agree for all problems. Invariant tests pass. |
| Structure | Computed by code | Page layout and problem count are computed by code. |
| Area | Level | What is checked |
|---|---|---|
| Answers | Computed twice and compared | Answers are computed by code from the problem structure. An independent AI solver reached the same answer without seeing the code answer. |
| Drawings | Computed by code | Bar models are drawn by code from the problem structure. |
| Wording | AI-written, independently checked | Story language is verified by an independent AI pass for naturalness and grade-appropriateness. |
| Area | Level | What is checked |
|---|---|---|
| Answers | Computed twice and compared | A round-trip test confirms the recorded corrections reproduce the clean text. An independent verifier fixed each corrupted sentence to the same text. |
| Wording | AI-written, independently checked | The clean (corrected) sentences are verified by an independent AI pass. |
| Area | Level | What is checked |
|---|---|---|
| Structure | Computed by code | Boxes, lines, checklist, and rubric all come from code. A readability estimate is computed by code. |
| Wording | AI-written, independently checked | Prompts and sentence starters are AI-written and verified by an independent AI pass. |
| Area | Level | What is checked |
|---|---|---|
| Data | Computed by code | All graph data values are generated by code from a stored seed. |
| Answers | Computed twice and compared | Every answer is computed by code, then re-computed with a second independent implementation. Both agree. An independent solver confirms. |
| Drawings | Computed twice and compared | Value-to-position inversion tests confirm that reading a drawn position back returns the exact value. |
| Wording | AI-written, independently checked | Scenario labels and realism are checked by an independent AI pass. |
| Area | Level | What is checked |
|---|---|---|
| Answers | Computed twice and compared | Every reading is computed by code, then re-computed with a second independent method. Both agree. |
| Drawings | Computed twice and compared | Inverse mapping tests confirm that reading a drawn position back returns the exact value. |
| Structure | Computed by code | Object captions and measurement ranges are checked for plausibility by code. |
TeachSpark uses cookies and local storage to improve your experience and process payments securely. Learn More