AI Projects for Science Teachers

You are an experienced science department chair reviewing a colleague's course, and I want the real problems rather than encouragement.

Here is my full year: [paste your unit list, your labs, and your major assessments].
Course and grade level: [Ex. 9th grade biology]. My standards are [NGSS, or your state framework].

Do four things, in this order:
1. List every performance expectation I hit three or more times, and every one I never touch at all.
2. Tell me which science and engineering practices students are actually doing, and which ones only appear as a phrase in my plans.
3. Do the same for the cross-cutting concepts, and name the ones that never show up in my year at all.
4. Give me the three highest-value changes I could make without rewriting the course, ranked by impact, with the cost of each.

Ask me anything you need before you start. Be direct, and tell me where you are guessing about what a unit actually covers.

I am building a unit around a phenomenon rather than a chapter.

Standard or performance expectation: [paste it].
Course and grade level: [Ex. middle school physical science].
Equipment and space I have: [Ex. one demo table, no fume hood, a class set of phones].

Give me five candidate phenomena. For each one:
1. What students see or experience, written the way I would present it on day one.
2. Why it is genuinely puzzling: what students' existing ideas would predict, and why that prediction fails.
3. The specific part of the standard a student cannot explain the phenomenon without.
4. How I would show it in my room (demo, video, data set, first-hand observation), what it costs, and how long it takes.
5. Any safety consideration, named plainly. Do not tell me a demo is safe. Name the hazards so I can check them against my SDS binder and my department protocols before I run it.

Then rank the five by how much argument they will produce in a real class, and say which one you would drop.

Act as a science curriculum designer who works in storylines rather than chapters.

Anchoring phenomenon: [paste it, or describe it].
Standards for the unit: [paste them].
Course and grade level: [grade]. The unit runs [number] class periods of [number] minutes.
Lessons or labs I already have and want to keep: [list them].

Sequence the unit so it builds instead of marching:
1. Order the lessons so each one ends on the question the next one takes up. Show that chain explicitly: for every lesson, the question students leave with.
2. Mark where students need to gather data before they can reason, so the lab comes before the explanation rather than after it.
3. Name the two places where students will want to jump ahead to the answer, and what to do about it.
4. Tell me where my existing activities fit, and which of them are now in the wrong place.
5. Flag the lesson where the sequence is weakest.

Do not write the lessons yet. I want the arc and the question chain first.

I need a driving question for a science unit.

Unit topic: [topic].
Standards: [paste them].
Course and grade level: [grade].
Anchoring phenomenon, if I have one: [describe it].

Give me five candidates. For each:
1. The question, worded exactly as it would go on the board.
2. What a student would have to gather, figure out, or reason through to answer it.
3. Whether a single search answers it, and say so plainly if it does.

Rules:
- Every question needs evidence and reasoning, not recall.
- Write them in student language for [grade level], not standards language.
- Nothing that reads as a lesson objective with a question mark on the end.

Then pick the strongest one, say why, and rewrite it two more ways so I can choose the phrasing my students will actually latch onto.

My lab is not happening. Give me a version that keeps the thinking.

The lab as I normally run it: [paste the procedure].
What students are supposed to learn from it: [the concept and the reasoning, not just the topic].
Why it is off: [Ex. equipment broken, class is remote, materials never arrived].
Course and grade level: [grade]. I have [number] minutes.

Give me two alternatives:
1. A data-analysis version, with a realistic data set students work from. Include believable scatter and at least one messy value, not a clean line.
2. A simulation or model version. Name the specific free simulation and where to find it. If you are not certain it exists and is still available, say so instead of guessing.

For each: the student handout ready to copy, the analysis questions, and the answer key.

Then tell me plainly what students lose by not touching the equipment, so I know what to circle back to later.

Write a substitute plan for one [number] minute period of [course and grade level] science.

Context: we are in the middle of [unit], students have already [what they have done], and the sub has no science background and has never met this class.

The plan must:
- Be a real lesson that moves the unit forward. Not a worksheet, not a video.
- Be a data-analysis or claim-evidence-reasoning task on our current content.
- Involve no lab work, no equipment, no chemicals, and nothing the sub has to supervise for safety.
- Require nothing from the sub except reading directions aloud and circulating.
- Produce something I can collect and actually use the next day.

Give me: a timed agenda, the exact words the sub should say to open class, the student handout ready to copy, the answer key, and two sentences on what to do if the room gets loud.

Build me a bank of 30 bell-ringers for [course and grade level] science, each designed to take five minutes at the start of class.

Our current unit is [unit], and the skills I want to keep warm are [Ex. energy transfer, graph reading, unit conversion, claims and evidence].

Rules:
- Every one connects to our current content or to a skill this course keeps needing. No generic trivia.
- Vary the type across the bank: a phenomenon to explain, a graph to read, a "what would happen if" prediction, a wrong answer to diagnose, a calculation or unit check.
- Each must be something students can start without me explaining it.
- None should take more than five minutes to review afterward.
- Nothing that requires equipment or any hands-on step.

Format as a numbered list. For each: the prompt exactly as students will see it, then the answer or the acceptable range, then a five-word note on what it practices.

Turn this cookbook lab into an investigation students design.

The procedure as written: [paste it].
Course and grade level: [grade].
Equipment and materials available: [list exactly what I have].
Time: [number] minutes across [number] periods.

Convert it:
1. Write the investigable question this procedure secretly answers, worded so students do not already know the outcome.
2. Tell me what to withhold and what students still need handed to them, so the task is open but not impossible.
3. Write the planning page students fill in before they touch anything: variables, what they will measure, how many trials, and what would count as an answer.
4. Give me the three procedures students are most likely to propose, including the two that will not work, and the question I should ask instead of correcting them.
5. List every safety issue that opens up once students are designing rather than following steps, including anything a plausible student plan could get wrong. State plainly that I have to check the final student procedures against my SDS binder and department protocols before anyone runs them, and do not approve any student design yourself.

I need to run this lab with different materials than it calls for, and I will be verifying everything you tell me before anything gets used.

The lab as written: [paste the procedure and materials list].
What I actually have: [list your inventory, with concentrations and quantities].
What I do not have: [list it].
Course and grade level: [grade].

For each missing item, give me candidate substitutions. For every candidate:
1. What it changes about the results, the timing, or the reasoning students do.
2. Every hazard it introduces or removes, named specifically, including anything it must never be combined with or heated near.
3. Exactly what I need to check before using it: which SDS entries, which department or district protocol, and which reference.
4. Your confidence, stated plainly, and say so directly where you do not know.

Rules:
- Do not tell me a substitution is safe or approved. You are listing candidates and hazards. I am the one verifying them.
- Never suggest changing a concentration, a quantity, or a heating step without saying outright that it has to be verified against a real safety source first.
- If a substitution is a genuinely bad idea, lead with that instead of listing it neutrally.

End with the substitutions you would not attempt at all, and why.

Write the pre-lab page for this investigation.

The lab: [paste the procedure].
Course and grade level: [grade].
What students already know: [list it].

The page has to make students think before they touch anything. Include:
1. Two prediction questions, each asking what will happen and why, so a guess looks different from reasoning.
2. Variable identification: independent, dependent, and the two controls students in this course most often miss.
3. A hypothesis frame that requires a mechanism, not just a direction.
4. One question about the measurement itself: what could go wrong in taking it, and how they would know.
5. Two safety questions specific to this procedure, written so a student has to name the hazard and the response rather than repeat a rule.

Rules:
- Every question specific to this lab. Nothing that could be asked of any lab.
- One page maximum, at a reading level my students can use.
- Flag any hazard you notice in my procedure, and remind me at the end to check the safety items against my SDS binder and department protocols before I hand this out.

Give me the answer key separately.

Write a five-question safety quiz for this specific lab.

The procedure: [paste it, including chemicals, concentrations, quantities, and equipment].
Course and grade level: [grade].
Safety equipment in my room: [Ex. goggles, aprons, eyewash, fume hood, fire blanket].

Requirements:
1. Every question about a hazard in this procedure, not general lab rules.
2. Each one asks the student to name a hazard and the correct response, or to spot what is wrong in a described action.
3. One question about what to do when something goes wrong in this particular lab.
4. Mix the formats: two multiple choice, two short answer, one scenario.

Then give me:
- The answer key.
- A separate list of every hazard you identified in my procedure, so I can compare it against my own review.
- A one-line sign-off students initial, which I keep as documentation.

Say clearly at the top of the answer key that your hazard list is a reading of the procedure, not a clearance, and that I have to verify it against the SDS for each chemical and my department protocols before I teach it.

Build the data table and graph template for this lab.

The procedure: [paste it].
What students measure, and with what instrument: [Ex. temperature with a digital probe to 0.1 C].
Number of trials: [number]. Course and grade level: [grade].

Give me:
1. A data table with correct column headings, units in the headings rather than in every cell, a row per trial, and a place to record instrument precision or uncertainty.
2. Space for the qualitative observations students always need somewhere to put.
3. A blank graph frame: the right variable on each axis, units labeled, a sensible scale for the range we should get, and a title format.
4. A short note telling students which graph type this data calls for and why, plus whether a line of best fit belongs here at all.
5. The two places students in this course most often mislabel something, and a one-line cue in the template that prevents it.

Format it so I can paste it into a document and print it. Keep it to one page.

Generate realistic sample data for this lab so my students can still do the analysis.

The procedure: [paste it].
What is measured, the instrument, and its precision: [describe it].
Expected relationship: [Ex. inverse, linear through the origin, exponential decay].
Range of the independent variable: [range]. Number of trials: [number].
Course and grade level: [grade].

Rules for the data:
- Believable scatter consistent with the instrument's precision. Not a perfect line.
- One or two values off enough that students notice and have to decide what to do with them. Tell me in the key which ones those are.
- Values in the real units and to the number of decimal places that instrument would actually give.
- Do not label the student copy as fabricated, but give me a line I can say to the class about where the numbers came from.

Then give me the analysis questions, the answer key with the calculated result, and the percent difference from the accepted value so students have something to compare against. Show your arithmetic so I can check it.

Write the error-analysis section for this lab.

The procedure: [paste it].
What students measured and with what: [describe it].
Our class results: [paste them, or say what went wrong].
Course and grade level: [grade].

Give me:
1. Five questions about the actual sources of error in this procedure, each naming the specific step where it enters.
2. For each, whether it is random or systematic, and which direction it pushes the result.
3. One question asking students to say how much a named error could account for, with numbers, so the answer is not just "human error."
4. One question about what they would change in the procedure, requiring a reason tied to a specific measurement.
5. The vague answers students give here ("we were not careful," "the equipment was old") and the follow-up question that pushes each one into something real.

Give me the answer key separately. Keep the student questions to one page.

I am about to teach [topic] to [course and grade level] students. Before I plan anything, tell me what students at this age already believe about it.

Give me:
1. The eight most common misconceptions, each written in a student's own words rather than as a correction.
2. Where each one comes from: everyday experience, earlier instruction, ordinary language, or a textbook figure that misleads.
3. Why each one is durable, meaning what evidence a student thinks supports it.
4. The specific observation, demonstration, or data that makes each one fail. Where that involves equipment or chemicals, name the hazards and say plainly that I have to check it against my SDS binder and department protocols before running it.
5. Two diagnostic questions for day one that surface these beliefs, worded so students answer honestly instead of guessing what I want.

Rank the eight by how much trouble each causes later in the unit, and tell me which one I have to deal with before anything else will stick.

Explain [concept] to [course and grade level] students five different ways.

What they already know: [list it]. What has already confused them: [describe it].

Give me:
1. A verbal explanation in plain language, using no term I have not taught.
2. A visual one: describe exactly what I draw on the board, in the order I draw it.
3. A mathematical one, with the relationship and what each symbol means physically.
4. A mechanical or hands-on one: what students do or manipulate to feel the idea.
5. An explanation by analogy.

Then, for the analogy and for every comparison anywhere in the other four, name what it gets wrong: which part does not map, and the misconception a student could walk away with because of it. Do this for every analogy, including the ones you slip in casually without calling them analogies.

End by telling me which of the five to lead with for this group, and which one to skip because the misconception it creates costs more than the clarity is worth.

Build a model-drawing task on [concept] for [course and grade level] students.

What students have observed so far: [describe it].
What they need to be able to explain: [the mechanism].

Give me:
1. The prompt exactly as students will see it, asking them to draw the mechanism and label what is happening, not just the parts.
2. What a correct model includes: the components, the interactions, what is happening at a scale they cannot see, and what the arrows have to mean.
3. Two common wrong models, described the way a real student would draw them, and what each one reveals about their thinking.
4. Three questions I ask a student while looking at their drawing, to find out whether the model means what it looks like it means.
5. A short student-facing checklist so they revise their own model once before I see it.

Keep the prompt short enough to sit at the top of a blank page. I want most of the page left for drawing.

Write a concept cartoon for [course and grade level] science on [concept or phenomenon].

The situation students are looking at: [describe the phenomenon].

Give me four characters, each making a different claim in one or two sentences of ordinary student speech:
- One correct.
- Three wrong in ways that come from real, common misconceptions, not carelessness.

Rules:
- Every claim has to sound reasonable. If the wrong ones are obviously wrong, the argument is over before it starts.
- No character hedges or gives a partial answer. Each one commits.
- Plain student language, no vocabulary I would test on.

Then give me:
1. Which misconception each wrong character represents.
2. The one question to ask after students have picked, to push them from opinion into evidence.
3. The observation or data that settles it, and what to do if the room lands on a wrong one and will not move.

Write the four claims as text I can drop straight into speech bubbles.

I am teaching [topic] and it is outside my training. My background is [Ex. biology, and my physics ended at one intro course fifteen years ago].

Course and grade level I am teaching it to: [grade].
What the standards ask students to do with it: [paste the performance expectation].

Bring me up to speed:
1. Explain the content one level deeper than my students will go, written for an adult who knows science generally but not this.
2. Name the vocabulary and notation this field uses that I am likely to use wrong.
3. List the student questions on this topic that catch a non-specialist out, with the answer to each.
4. Tell me where a teacher from my background typically gets this subtly wrong.
5. Point at the specific claims in your own explanation I should verify before teaching, and where to verify them. Say plainly which parts you are less sure about.

No reassurance. I would rather know where the thin ice is.

I want to connect [unit or topic] to current research for my [course and grade level] students.

Suggest [number] recent findings or lines of work that genuinely connect to what we are studying.

For each one:
1. The finding, in two sentences a student at this level can follow.
2. Which part of our unit it connects to, and the question it lets me ask that the textbook does not.
3. The citation: authors, title, journal or outlet, and year, complete enough that I can search for it and find it.
4. Your confidence that this work exists as you have described it, stated plainly.

Rules:
- I am looking every one of these up before I say it in class. If you are not certain something is real, say so instead of producing a plausible citation. An honest "I am not sure this exists" is worth more to me than a clean reference.
- Prefer well-covered findings I can reach through a news article or a press release when the paper is paywalled.
- Do not invent anything to fill out the list. A short list of real work beats a long one I have to disprove.

Then tell me which is worth ten minutes of class and which are only worth a mention.

Build claim-evidence-reasoning scaffolds for this specific investigation.

The lab or data set: [paste the procedure and the results students will have].
The question students are answering: [paste it].
Course and grade level: [grade].

Give me:
1. Sentence frames for each part, written so the frame does not hand over the content. Include the reasoning frame students most need, the one that forces them to connect their evidence to a principle.
2. A model response at the level I want, using our actual data, annotated to show which sentence does which job.
3. A weak response for students to improve: a real claim, evidence that is described rather than used, and reasoning that only restates the claim.
4. The answer key for the weak one: what is missing, and what a good revision adds.
5. A three-item checklist a student runs on their own paragraph before turning it in.

Keep it to two pages, and use our data and our vocabulary rather than a generic example.

You are giving feedback on one student lab report. Read my rubric first, then the report.

Rubric: [paste it].
The lab and the question students answered: [paste it].
Course and grade level: [grade].
Report: [paste the report with the student's name removed].

Respond with exactly this and nothing else:
1. Two specific strengths, each quoting the sentence, calculation, or graph choice that earned it.
2. The single revision that would most improve this report, with one concrete example of what it would look like in this student's own data.
3. One question I should ask this student about their reasoning.

Rules: address the work, never the student. Do not rewrite the report. Do not give a grade or a score. If a calculation or a unit is wrong, say so and show the correct version, but flag it as something I need to check myself. Under 150 words, at a reading level this student can use.

Write three model lab reports for [course and grade level] science: one at the top of my rubric, one solidly in the middle, and one that is a common near-miss.

The lab: [paste the procedure].
The data students will have: [paste it, or describe the results].
Rubric: [paste it]. Length: about [number] words each.

Then annotate all three. After each one, list the specific moves that put it at that level, in my rubric's own language. For the near-miss, name exactly what it is doing that looks like scientific reasoning but is not, especially where it describes the data instead of using it.

Rules:
- All three use the same data, so the difference is the reasoning and nothing else.
- Make them sound like real students at this grade, including the small awkwardness real writing has.
- Keep the arithmetic and the units correct in all three, and tell me at the end which numbers I should check before I hand these out.

I want my [course and grade level] students arguing from evidence rather than opinion.

Our unit is [topic].

Give me [number] scientific questions connected to this unit that were genuinely open at some point, meaning competent scientists disagreed and the evidence available then supported more than one reading.

For each:
1. The question as it stood then, and roughly when.
2. The evidence on each side, specifically, at a level [grade] students can work with.
3. What eventually settled it, kept in a separate section so I can withhold it.
4. The claim each group defends, and the single piece of evidence hardest for each side to explain.

Rules:
- Use questions that were actually contested among scientists. Do not hand me a settled question dressed up as a debate, and do not use a topic where the science is clear and only the public conversation is contested.
- Say plainly if you are unsure a disagreement went the way you describe, and give me enough detail to check it.

End with the one that will produce the best argument in a real classroom, and why.

Build a data-literacy set for [course and grade level] science, connected to our unit on [topic].

Give me [number] graphs or data displays. Describe each one precisely enough that I can recreate it or find it: the variables, the axes, the scale, the units, the shape of the data, and the source if it is real.

Include at least three that are misleading in a specific, nameable way: [Ex. a truncated axis, a scale that changes mid-axis, a correlation presented as a cause, a percentage with no base, a cherry-picked range].

For each display, write:
1. Two questions about what it actually shows.
2. One question about what it does not show, or cannot.
3. One question about a choice the person who made it had to make.

Rules:
- Do not label the misleading ones on the student copy.
- If you cite a real graph or data set, give me enough detail to find it, and say plainly if you are not certain it exists.

Give me the answer key separately, naming the specific distortion in each misleading display.

Write a confident 400-word explanation of [topic] for [course and grade level] students.

Include at least eight specific factual claims: numbers, dates, named mechanisms, quantities, and comparisons.

Write it entirely from memory. Do not look anything up, do not hedge, and do not flag anything as uncertain. Use the assured voice of a textbook sidebar.

I am asking for this deliberately. My students are going to check every claim against our textbook, a reference source, and their own data, and find what you got wrong. So do not correct yourself, do not soften any claim, and do not add a disclaimer at the end.

One limit: include no procedure, quantity, or chemical combination a student could go try. This is a reading task only.

After my students have finished, I will come back and ask you to grade your own accuracy, claim by claim.

I am teaching my [course and grade level] students that AI invents sources, and I want them to find that out themselves rather than hear it from me.

Give me [number] sources on [topic]: papers, articles, or books, each with authors, title, journal or publisher, year, and a one-sentence summary of the finding.

Write them from memory, in the confident format of a bibliography. Do not mark any of them as uncertain and do not add a disclaimer, because the point of the lesson is that students cannot tell which is which by looking.

Then, in a separate section I keep to myself, tell me for each one: whether you believe it is real, how confident you are, and what specifically you may have gotten wrong (the author, the year, the journal, the finding, or the whole thing).

Also give me the four search steps students should take before deciding a source does not exist, so a real paper sitting behind a paywall does not get called fake.

My [course and grade level] students are comparing three explanations of [concept].

First, write your own explanation, about [number] words, at the level a student would find on a homework help site. Write it plainly and do not hedge.

Then help me set up the comparison:
1. What to look for in our textbook's version, and the specific questions to ask about what it simplifies.
2. What a working scientist would emphasize that neither of the other two does, and where I might find one saying it (a lecture, an interview, a review article). Say plainly if you are not certain a specific source exists.
3. A comparison table students fill in: what each source gets right, what it leaves out, who it is written for, and what it assumes the reader already knows.
4. The three questions I ask at the end, including one about why the three differ rather than which is best.

Then tell me the weakest part of your own explanation above, specifically.

Draft a family letter and permission form for an activity that needs one.

The activity: [Ex. fetal pig dissection, a lab using dilute hydrochloric acid, a stream study off campus].
Course and grade level: [grade].
When and where: [date and location].
Safety equipment and supervision: [describe it].
Our alternative for a student who opts out: [describe it, or ask me for one].

Give me:
1. The letter: what students will do, what they learn from doing it, and why this activity rather than a substitute. Under 300 words, no jargon.
2. The safety plan in plain language: the hazards, the precautions, the supervision, and what happens if something goes wrong.
3. The opt-out paragraph, written so a family does not feel it is asking for a favor.
4. The permission form itself, with lines to sign and date.
5. The three questions a family is most likely to ask, with a short answer to each.

Then list what I have to confirm before this goes out: which district or department policy, which SDS entries, and what my administrator needs to approve. This is a draft for review, not a cleared document.

Two things, in this order.

First, a progress-report comment bank for [course and grade level] science. The skills I report on are [Ex. investigation design, data analysis, scientific explanation, lab practice and safety, participation in argument].

For each skill, write five comments at each level: exceeding, meeting, approaching, not yet meeting.
Rules: name a specific skill or behavior, never effort or personality. Include a concrete next step in each. Use [NAME] as a placeholder so I can drop in a name and pronoun. Two sentences maximum. Nothing that would embarrass a student reading it over a parent's shoulder.

Second, a funding request for equipment I need: [what it is, the cost, and the vendor if I have one].

Write it so a non-scientist understands what students will be able to investigate that they cannot investigate now. Include what students will do with it, which standards it serves, how many students it reaches per year, and how long it will last. Under 400 words, no jargon, and no line about being underfunded.

Ask me for anything you need before you write the request.