About 30% of higher ed instructors used generative AI daily or weekly in spring 2025, up from just 4% in spring 2023.Source ↗
Curriculum development made up 57% of professors' AI conversations in Anthropic's analysis of ~74,000 higher ed chats, ahead of academic research (13%) and assessing student work (7%).Source ↗
86% of faculty see themselves using AI in teaching in the future, but 80% say their institution hasn't made clear how AI should be applied in teaching.Source ↗
38% of instructors report AI has increased their workload — mainly monitoring for cheating (71%) and redesigning assessments (61%) — and only 28% of institutions have a formal AI policy.Source ↗
planningClaudeChatGPTGemini

Planning a class session that isn't 75 minutes of straight lecture

Curriculum development is the single biggest way professors use AI — 57% of educators' AI conversations in Anthropic's sample. The real pain point isn't content knowledge; it's converting a topic you know cold into a timed session with activities, discussion structure, and checks for understanding, the night before class and on top of research and service obligations.

Prompt
You are an experienced higher-education instructional designer who specializes in
active learning. Design a complete {{session_length}}-minute class session for my
course, {{course}}, on the topic of {{topic}}.

Student context: {{student_context}}

Output format:
1. One measurable learning objective ("By the end of this session, students will be able to...")
2. A 5-minute opener that surfaces prior knowledge or a common misconception
3. Two short lecture segments (10-12 minutes each): key points plus one worked
   example or short case for each
4. One active-learning block (15-20 minutes): think-pair-share, small-group
   problem, or structured debate — with exact instructions I can read aloud and
   what to listen for while circulating
5. A closing check for understanding: a one-minute-paper prompt or two poll
   questions with answers
6. A timing table and one "plan B" cut if discussion runs long

Constraints: build only on the concepts and readings I named — do not assign,
cite, or paraphrase readings or sources I did not mention, and do not invent
facts or statistics for the lecture segments. Keep every timing realistic. If the
topic cannot be covered well in this session length, say so and propose what to
move to another session.

Fill in your details and the prompt updates live — then copy.

automationChatGPTClaude

Writing exam questions from your own lecture notes

Among instructors who say AI increased their workload, 61% point to redesigning assessments to counter student AI use — which in practice means writing fresh question banks and multiple versions every term. Forcing the model to work only from your materials, with distractors built on real misconceptions, turns that grind into an editing job.

Prompt
You are a university {{discipline}} instructor writing a fair, rigorous exam.
Using only the course material below — do not add outside facts, formulas, or
examples — create exam questions.

Course material: {{source_material}}

Question spec: {{question_mix}}

Requirements:
- Label each question with the concept it tests and its cognitive level
  (remember, apply, or analyze).
- Multiple choice: four options with one clearly correct answer. Build each
  distractor around a specific misconception students actually hold, and note
  that misconception in brackets after the distractor.
- Short answer: answerable in 2-4 sentences, with a model answer and a
  point-by-point scoring note.
- Order questions from recall to application. No trick questions, no "all of
  the above," no negatively worded stems.
- Then produce a second version of the whole exam — reworded stems, shuffled
  options, same concepts and difficulty — for make-ups or a second section.
- End with a separate answer key for both versions.

If the material I provided cannot support the number or level of questions I
asked for, tell me exactly what is missing instead of padding with outside
knowledge.

Fill in your details and the prompt updates live — then copy.

analysisClaudeChatGPT

Building a rubric and calibrating grading with your TAs

Faculty rate grading as the task AI does worst, yet nearly half of grading-related AI conversations hand the judgment over anyway — the exact pattern researchers called concerning. The defensible version: use AI before grading starts, to build the rubric, the comment bank, and a calibration packet that gets you and your TAs scoring consistently.

Prompt
You are an assessment designer for university courses. Create a grading kit for
this assignment.

Assignment: {{assignment_description}}
Course and level: {{course_level}}
Skills to assess: {{focus_skills}}

Output three parts:
1. A four-level analytic rubric (Excellent / Proficient / Developing / Beginning)
   with one row per skill. Descriptors must describe observable features of the
   work in 25 words or fewer per cell, with no overlapping language between
   adjacent levels and no vague qualifiers like "adequate."
2. A feedback comment bank: for each skill, two sentence starters for work that
   meets the standard and two for work below it. Every below-standard comment
   must name one concrete action the student can take in revision.
3. A calibration packet for my TAs: write three short synthetic response
   excerpts — invented by you, not real student work — at clearly different
   quality levels, each with a suggested rubric score and a one-line rationale,
   so we can norm our grading before we start.

Constraints: assess only the skills I listed — do not add criteria. Keep all
rubric language student-readable, because the rubric will be published with the
assignment. Do not ask me for real student submissions.

Fill in your details and the prompt updates live — then copy.

writingClaudeChatGPT

Red-teaming a grant proposal before reviewers see it

Grant writing is one of the most collaborative faculty AI uses — 70% augmentation rather than automation in Anthropic's data — and a Wiley survey of nearly 5,000 researchers found most expect AI in grant writing to become widely accepted. You cannot use AI to review other people's proposals, but you can make it the harshest reader your own draft meets before submission.

Prompt
You are a skeptical, experienced grant reviewer for {{funder}} — the kind who
reads fifteen proposals in a weekend and rewards clarity over ambition. Review
the draft section below against the funder's published criteria.

Draft section: {{draft_text}}

Review criteria, pasted from the funder's official guidance: {{review_criteria}}

My biggest worry about this proposal: {{main_concern}}

Output:
1. A reviewer-style summary: three genuine strengths, then weaknesses ranked by
   how much each would hurt the score
2. Line-level flags: quote every sentence where a claim outruns the evidence
   presented, where aims appear dependent on each other's success, or where
   jargon would slow a panelist outside my subfield
3. The five questions a reviewer would most likely raise in panel discussion
4. Suggested rewrites for the three weakest passages only, keeping my voice and
   technical content

Constraints: do not invent citations, preliminary data, methods, or results —
where support is missing, flag the gap rather than filling it. Judge only
against the criteria I pasted, not criteria you recall from memory. Do not
soften criticism to be encouraging; I need the panel's objections now, not in
the summary statement.

Fill in your details and the prompt updates live — then copy.

communicationClaudeCopilot

Drafting recommendation letters that stay honest and specific

Every fall brings a stack of recommendation letters, each needing an hour you don't have — and Anthropic's data shows drafting them is one of the admin tasks professors most often hand to AI. Done carelessly, that means template letters and student data in a consumer tool. Done right, it means your notes, your judgment, and a draft that saves the composition time.

Prompt
You are helping a professor draft a letter of recommendation. Privacy rules for
this chat: I will refer to the person only as "the student" — never ask for a
name — and you will write the letter with [Name] placeholders that I will fill
in later in my word processor.

My notes on the student: {{student_notes}}
They are applying to: {{target_program}}
How I know them: {{relationship_context}}
Endorsement strength, honestly stated: {{endorsement_level}}

Structure:
1. Opening: who I am, how long and in what capacity I have known [Name], and a
   one-sentence endorsement that matches the strength I stated
2. Two body paragraphs, each built around one concrete example from my notes —
   evidence, not adjectives
3. A closing that connects [Name]'s record to this specific program and offers
   contact for follow-up

Constraints: use only facts from my notes — do not invent projects, grades,
rankings, or personality traits. No stock phrases ("It is my distinct
pleasure..."). Do not escalate my stated endorsement level; a credible letter
serves the student better than an inflated one. If my notes are too thin to
support a persuasive letter, reply "NEED MORE" and ask me three specific
questions instead of padding.

Fill in your details and the prompt updates live — then copy.

creativeClaudeChatGPT

Building an interactive simulation for a concept students keep missing

In Anthropic's report, professors used AI to build custom teaching tools that used to be prohibitively expensive in time — chemistry stoichiometry practice tools, physics models, and review games — often as single-file web apps. If one concept defeats a chunk of your class every year, an interactive model students can poke at is now an evening's work instead of a summer project.

Prompt
You are a developer-educator who builds small, self-contained teaching tools for
university courses. Build an interactive simulation as a single HTML file with
inline CSS and JavaScript — no external libraries, fonts, or network calls — that
I can upload to my LMS or project in class.

Course and level: {{course_level}}
Concept to teach: {{concept}}
What students should do with it: {{learner_task}}
Source content to build from (use only this): {{content_source}}

Requirements:
1. Controls (sliders, inputs, or buttons) for the two or three parameters that
   matter most, with instant visual feedback
2. A live numeric readout showing the quantitative relationship, not just an
   animation
3. A "challenge mode" with five questions drawn strictly from my source content,
   giving hints — not answers — on wrong attempts
4. On-screen instructions a first-time user can follow without my help
5. Legible at classroom-projector size and usable on a laptop trackpad

Constraints: stay accurate to the source content I provided — do not import
formulas, constants, or claims from elsewhere. If you must simplify the model to
make it interactive, list every simplification in an HTML comment at the top of
the file so I can review it before class.

Fill in your details and the prompt updates live — then copy.

Common questions from professors

Am I allowed to use AI to grade student work?

Institution policies vary and most are still unwritten — only 28% of institutions had a formal AI policy as of spring 2025. The stronger caution is practical, since faculty themselves rate grading as the task AI does worst. Use AI for rubric language and comment drafting, keep every scoring decision yourself, and never paste identifiable student work into a consumer tool.

Does pasting student work into ChatGPT violate FERPA?

It can. Student submissions with names or identifying details are education records, and consumer AI tools have no data agreement with your institution. Strip names, IDs, and headers before pasting anything, or use an institution-licensed tool covered by a data-protection agreement. When in doubt, describe the work instead of pasting it.

Can I use AI to help review manuscripts or grant proposals?

For federal peer review, no — NIH prohibits generative AI in its review process, and NSF treats uploading proposal content to non-approved tools as a confidentiality violation. Journal policies vary but most major publishers restrict it for the same reason. Confidential material you're reviewing should never enter a tool you don't control.

Do I need to tell students when I use AI to prepare course materials?

Increasingly, yes — at least as a norm. In one widely covered 2025 case, a Northeastern student demanded a tuition refund after discovering a professor's AI-generated lecture notes. If your syllabus sets AI rules for students, a short statement about your own use is cheap insurance and models the transparency you're asking of them.

Related professions