Teachers are among the most time-poor professionals in any workforce. The visible part of teaching, the time in front of students, represents a fraction of the total work: lesson planning, assessment design, marking, parent communication, administrative reporting, resource creation, professional development, and the informal pastoral support that sustains struggling students. The average teacher works significantly more hours than their contracted time suggests.
AI has arrived in education not primarily as a classroom technology but as a professional tool that teachers are using to manage the invisible workload that surrounds the teaching itself. This is the story that gets less attention than the debate about whether students are using AI to cheat, and it is arguably more consequential for the quality of education.
This
post covers what teachers are actually doing with AI in 2026: the specific uses
that are saving meaningful time, the uses that are generating genuine concern,
the debates the profession is having with itself, and the ways AI is beginning
to change what happens in the classroom itself rather than just what surrounds
it.
The Invisible Workload: Where AI Is Saving Teachers the Most Time
Lesson
Planning and Resource Creation
Adaeze, the secondary school teacher in Enugu introduced in the Day 6 post, was spending most of every Sunday writing lesson plans, assessment rubrics, and parent communication letters. She is not unusual. Lesson planning to a high standard is time-consuming, and the expectation that teachers will differentiate their planning for students at different ability levels multiplies the workload significantly.
AI tools, particularly Claude and purpose-built education tools like MagicSchool AI, are being used by teachers to generate lesson plan frameworks, differentiated activity options for different ability levels, discussion questions at varying depths of complexity, assessment rubrics, and starter activities. The key workflow that consistently works: the teacher provides the learning objective, the year group or grade level, the ability range of the class, and any specific curriculum requirements. The AI produces a structured starting point that the teacher then adapts with their specific knowledge of the class.
A
primary school teacher in Kumasi described this as reducing her Sunday planning
sessions from four hours to 90 minutes. The planning is still happening. The
mechanical production of the first draft is not. The time saving allows her to
put more thought into the adaptation rather than the generation.
Testing Note: When Claude was given a learning
objective (students will understand the causes of the First World War at a
secondary level), a class description (Year 9, mixed ability, including three
students with learning support needs), and asked for a 50-minute lesson plan
with differentiated activities, it produced a structured plan with three
activity tiers in under 60 seconds. The teacher reviewing the output estimated
it would have taken her 40 minutes to produce manually and required 15 minutes
of editing to match her actual classroom context.
Assessment
and Feedback
Marking is one of the most time-consuming and mentally draining parts of a teacher's workload. AI is beginning to assist in two ways: generating assessment rubrics and mark schemes, and providing first-pass feedback on student work.
Rubric generation is well-established and low-controversy. Give Claude a learning objective and an assessment task and ask for a detailed rubric with descriptors at four or five performance levels, and it produces a solid starting point in seconds. Teachers who previously spent an hour constructing rubrics are now spending fifteen minutes reviewing and refining AI-generated ones.
AI-assisted marking of student work is more contested. Several schools are piloting systems where student essays are processed through AI for initial feedback on structure, argument development, and writing quality, with teacher review before feedback is returned to students. The educational technology companies promoting these systems report efficiency gains. The teaching profession's unions and many individual teachers are concerned about the accuracy of AI feedback on nuanced work and about the message it sends to students about the value of their learning.
The
concern that deserves the most weight is not efficiency. It is the nature of
marking as a learning experience for teachers. Reading student work carefully
is how teachers understand what their students actually understand. Outsourcing
that reading to AI, even partially, risks breaking a feedback loop that is
central to effective teaching.
Parent
and Administrative Communication
Parent communication letters, progress reports, referral documentation, and administrative reports constitute a significant portion of teacher time that involves writing to formulaic conventions rather than genuine creative or pedagogical thinking. This category of writing is well-suited to AI assistance. A teacher who needs to write twelve progress reports can give Claude the student name, key observations, and the tone required and generate a draft for each in minutes rather than hours.
A
special educational needs coordinator in Manchester described using this
workflow to produce draft documentation for 28 annual review reports in an
afternoon, a task that previously consumed two weeks of evenings. The drafts
required individual review and personalisation, but the structure, the standard
phrasing, and the formulaic sections were handled by AI. The SEN coordinator's
time went to the parts requiring professional judgment about individual
students.
AI in the Classroom: What Is Actually Happening With Students
The debate about AI and student assessment, whether students are using AI to cheat and what schools should do about it, has been covered in the Day 9 post on this blog. This section focuses on something different: the ways teachers are deliberately introducing AI into learning activities as a teaching tool rather than prohibiting it as a cheating mechanism.
AI as a
Thinking Partner in the Classroom
Some teachers are using AI as a classroom thinking partner, asking students to interact with an AI chatbot as part of a structured activity and then critically evaluate the responses they receive. A history teacher might ask students to have Claude argue a historical position and then identify three flaws in the AI's argument. An English teacher might have students compare their own creative writing to an AI-generated piece on the same brief and articulate specifically what their writing does that the AI's does not.
These uses treat AI literacy as a curriculum objective in itself. Understanding how to evaluate AI outputs, identify hallucinations and biases, and use AI tools responsibly is a skill the profession is increasingly recognising as foundational for students who will enter a workforce where AI is ubiquitous.
Personalised
Practice at Scale
AI is being used in some classrooms to generate personalised practice questions for individual students based on their demonstrated gaps and learning pace. A mathematics teacher can give Claude a specific student's performance data and ask for ten practice questions at the appropriate difficulty level, targeting the specific concept the student has not yet mastered. This level of personalisation was previously only achievable by teachers working one-on-one, not in classes of thirty.
A
secondary school mathematics teacher in Johannesburg described using this
approach for homework differentiation. Rather than setting the same homework
for all students, she generates three difficulty-differentiated versions of
each homework set using Claude, targeted at her three broad ability groups. The
time to generate all three versions is approximately 20 minutes. The time to
mark and respond to differentiated homework has not increased because the
practice questions are calibrated to each group's current level, which means
fewer students are working on problems that are too easy or too hard.
The Concerns Teachers Are Raising
The teaching profession's engagement with AI is not uniformly enthusiastic, and the concerns being raised by teachers deserve honest engagement rather than dismissal as resistance to change.
The
Homogenisation of Teaching
If thousands of teachers use the same AI tool to generate lesson plans, are the lessons subtly converging toward a common template? Teaching is a creative professional practice, and the diversity of approaches, the teacher who uses drama to teach history, the one who connects mathematics to cooking, the one who uses their unusual background to make unexpected curriculum connections, is itself valuable. There is a genuine concern that AI-generated lesson frameworks could gradually narrow the range of pedagogical approaches rather than expanding them.
Data
Privacy and Student Information
Using AI tools for student-related work raises immediate data protection questions. Entering student names, learning difficulties, or assessment data into consumer AI tools almost certainly violates data protection requirements in most jurisdictions, regardless of how convenient the tools are. Schools and individual teachers need clear policies about which AI tools are approved for use with student data and which are not. The consumer free tiers of ChatGPT, Claude, and Gemini are not appropriate for inputting identifiable student information without specific institutional data processing agreements in place.
The Risk
of Over-Reliance
A
teacher who has always relied on their own curriculum knowledge and pedagogical
judgment to plan lessons develops that knowledge and judgment through the
practice of planning. A new teacher who relies heavily on AI-generated lesson
plans from the beginning of their career may develop more slowly in exactly the
professional capabilities that make teaching sustainable and effective long-term.
This is not a reason to prohibit AI use for new teachers. It is a reason to
build deliberate reflection practices around AI use rather than accepting AI
output uncritically.
The Schools Getting This Right
The schools and education systems navigating AI most effectively are the ones that have moved from prohibition to framework. They have made explicit decisions about which AI uses are encouraged for teachers, which require institutional approval, and which are not appropriate for student data contexts. They are treating AI literacy as a curriculum objective with the same seriousness they apply to digital literacy. And they are involving teachers in the design of AI policies rather than imposing them from above.
In
several West African countries, teacher professional development organisations
have begun incorporating AI literacy into continuing professional development
programmes, recognising that teachers who understand AI tools will be better
equipped to guide students in using them responsibly than teachers who have
been told only to watch for cheating.
The AI Vanguard Take:
The most
important AI story in education is not whether students are using AI to cheat.
It is whether teachers are being given the time, support, and professional
development to engage with AI as a genuine opportunity rather than managing it
purely as a threat. Teachers who understand AI well will produce students who
understand it well. That is where the educational investment in AI needs to
concentrate.
Frequently Asked Questions
Which AI
tools are best for teachers?
For general lesson planning, differentiation, and communication tasks, Claude is consistently rated highest by teachers for its instruction-following precision and writing quality. ChatGPT is more accessible for teachers new to AI. MagicSchool AI is a purpose-built education tool designed specifically for teaching contexts with built-in templates for common teacher tasks. Canva AI is valuable for creating visual resources and classroom displays. All have free tiers appropriate for general use.
Is it
appropriate to use AI-generated lesson plans without telling students?
Using AI as a tool in lesson planning is analogous to using any other planning resource and generally does not require disclosure to students. However, in contexts where teaching AI literacy is a curriculum goal, being open about your own AI use as a teacher can be a powerful modelling opportunity: it demonstrates responsible, critical use of AI tools and gives students a concrete example of how professionals engage with the technology.
How
should schools handle the question of students using AI?
The
Day 9 post on this blog covers this question in full from the student
perspective. From the institutional perspective: schools that have moved from
blanket prohibition, which is both unenforceable and educationally
unproductive, to disclosure-based frameworks with clear guidance on which uses
are acceptable for which assessment types are reporting better outcomes than
schools still in prohibition mode. The key is specificity: students need to
know not just that AI use is sometimes acceptable but exactly which uses are
acceptable for which types of work.
For Educators on The AI Vanguard: The AI Vanguard covers AI in education, AI ethics, and AI for
students regularly. Subscribe below to receive every post.
