Some Teachers Are Using Ai To Grade Their Students, Anthropic Finds - Why That Matters

1 week ago

Follow ZDNET: Add america arsenic a preferred source on Google.

ZDNET's cardinal takeaways

Anthropic published its Education Report, analyzing educators' Claude usage.
Teachers are utilizing Claude to thief people students, a arguable usage case.
AI companies are doubling down connected devices for education.

Much of nan attraction connected AI successful acquisition is connected really students will beryllium affected by AI tools. Many are concerned that nan enticement to cheat and AI's erosion of critical reasoning skills will diminish nan value of their education. However, Anthropic's latest acquisition study focuses connected educators' outlook connected AI successful nan schoolroom -- and finds immoderate astonishing ways teachers are implementing nan tech.

Also: The tasks assemblage students are utilizing Claude AI for most, according to Anthropic

AI companies are alert of nan tensions users acquisition betwixt utilizing AI arsenic a copilot aliases support and letting it automate definite parts of their work. Anthropic's study shows really educators are navigating that tension, and really those choices alteration connected a lawsuit by lawsuit basis.

To behaviour this report, Anthropic analyzed anonymized conversations betwixt Claude.ai, its chatbot, and Free and Pro accounts associated pinch higher acquisition email addresses, and filtered for education-specific tasks from May and June of 2025. Within that clip period, Anthropic identified 74,000 conversations involving tasks specified arsenic creating syllabi, grading assignments, and more.

The institution besides matched each speech to nan astir fitting task from nan database of acquisition tasks successful nan US Department of Labor's Occupational Information Network (O*NET) database. Separately, Anthropic besides said that it bolstered its study pinch study information and qualitative investigation from 22 Northeastern University module members who are early adopters of AI.

For a complete breakdown of nan setup, you tin publication nan report. Now for nan findings.

Some educators are automating grading

Anthropic recovered that nan astir communal usage cases of AI for educators were program improvement (57%) and world investigation (13%). In a smaller usage case, however, Anthropic recovered that nan remaining 7% of educators utilized Claude to "assess student performance," which includes giving students feedback, grading against rubrics, and summarizing evaluations -- contempt nan outlook galore teachers stock that grading is simply a less-wise usage of AI.

Also: AI agents get successful US classrooms

When educators utilized Claude to grade, they relied connected it to nan constituent of automation astir half nan clip -- 48.9%. "That's contempt pedagogue concerns astir automating appraisal tasks, arsenic good arsenic our surveyed module standing it arsenic nan area wherever they felt AI was slightest effective," Anthropic said.

By contrast, nan study showed teachers utilized AI to augment tasks for illustration school and instruction, penning assistance proposals, world advising, and supervising world work. Besides grading, tasks pinch higher automation tendencies included managing acquisition institution finances and fundraising, maintaining student records, and managing world admissions and enrollment -- galore of which tin beryllium much admin-heavy.

The shape from those choices shows that educators are much consenting to automate tedious, method tasks. However, for those who require much analyzable and captious thinking, educators will usage AI to collaborate instead.

Also: Where AI educators are replacing teachers - and really that'll work

Anthropic added that nan precocious percent suggesting automated grading is concerning -- fundamentally expressing siren astatine nan thought that educators are handing specified a delicate portion of school disconnected to AI. The interest is astatine slightest a partial acknowledgement that AI whitethorn not beryllium recommended for specified a task, and conveys immoderate deficiency of assurance connected Anthropic's portion that Claude should beryllium utilized this way.

One Northeastern professor Anthropic worked pinch agreed, citing ethical concerns and accuracy issues: "I person tried immoderate experiments wherever I had an LLM people papers, and they're simply not bully capable for me. And ethically, students are not paying tuition for nan LLM's time, they're paying for my time. It's my civilized responsibility to do a bully occupation (with nan assistance, perhaps, of LLMs)."

Even arsenic nan smallest usage case, Anthropic noted it was nan 2nd astir automated task.

"While it's not clear to what grade these AI-generated responses facet into nan last grades and feedback, nan interactions surfaced by our investigation do show immoderate magnitude of delegation to Claude," Anthropic wrote.

Other ways teachers usage AI

Other unsocial usage cases recovered successful nan information included creating mock ineligible scenarios for acquisition simulations, processing workforce training content, drafting proposal letters, and creating gathering agendas.

While nan Northeastern module reported utilizing AI for their ain learning arsenic different communal case, Claude.ai study was not capable to corroborate it because of challenges pinch nan filtering mechanism. The Northeastern module did propose that educators are leveraging AI for these tasks because AI tin automate tedious tasks, collaborate arsenic a thought partner, and personalize learning experiences for students, according to nan report.

Also: My apical 5 free AI devices for schoolhouse - and really they tin thief supercharge your learning

Beyond utilizing nan existing devices for schoolroom help, teachers are besides building their ain AI tools. For example, Anthropic said teachers are often utilizing its Artifacts feature, which allows users to create an app without coding, to build "interactive acquisition materials." These creations see interactive acquisition games, appraisal and information tools, information visualization, world calendars and scheduling tools, fund planning, and more.

AI's acquisition creep

Just successful clip for back-to-school season, AI companies person been on a tear to merchandise devices marketed toward some students and teachers. Anthropic recently launched a caller Learning Mode in Claude.ai chatbot and Claude Code, a complement to OpenAI's Study Mode -- some intend to employment nan Socratic method to create a back-and-forth pinch a personification alternatively than spitting retired answers. Elsewhere, text-to-speech app Speechify launched a competitor to NotebookLM's AI podcast tool, and Google made its $20-per-month suite of AI tools free to assemblage students.

Also: Why AI chatbots make bad teachers - and really teachers tin utilization that weakness

Putting nan statement astir AI's domiciled successful acquisition speech for a moment, a assemblage statement tin beryllium lucrative -- arsenic tin creating a student dependency connected your devices to get done a reliable semester. Given really burnt retired teachers are by definition, particularly fixed the COVID-19-related exodus from nan profession, is it really a astonishment that immoderate educators are changing their tune connected automating parts of their occupation pinch AI? Can an manufacture gunning to beryllium successful everyone's workflow, including successful nan classroom, definitive interest erstwhile that originates to happen, particularly without argumentation restricting definite uses of AI, aliases guidance from those companies themselves?

AI successful nan schoolroom is still excessively nascent to show wherever this will go, but for now, AI companies are wading into -- and creating -- a analyzable early for education. Individual schoolhouse and assemblage policies whitethorn yet find nan outcomes, and moreover then, pinch a constricted scope of control, arsenic these devices stay truthful easy accessible.