The Future of Teaching Skills: Why AI Makes Productive Struggle Your Job

You trained to explain. To take a difficult idea, break it into steps and walk a room of students through it until something clicks. For most of your career, that skill defined a good teacher.

Then a free app started doing it in seconds, at any hour, in any language, at whatever level a student asks for.

If that has left you quietly wondering where you fit, you are asking the right question. In a Suraasa webinar hosted by Manish Pandey, our co-founder and CTO Ankit Khanna gave a clear answer. AI has commoditised explanation. What it has not touched, and has actually made more valuable, is a teacher's ability to design the right kind of difficulty.

This article walks through his argument, the cognitive science behind it, and what it means for the future of teaching skills in your own classroom. Not in theory. On Monday morning.

The Flaws of Legacy Teacher Professional Development

Before Ankit looked at students, he looked at teachers. Specifically, at how schools have tried to help teachers grow.

To hear the full argument in his own words, including the data behind it, watch Ankit Khanna's session below.

You know the model. A one-day or two-day workshop. A guest speaker. A slide deck. A certificate of attendance that goes into a file. Then back to the classroom, where nothing structurally changes.

Ankit did not soften his verdict on it:

"The way teacher trainings were delivered a decade ago, still most of the people are delivering it, is the most expensive way to retain the least amount of information." (Ankit Khanna)

His diagnosis is precise. Schools built these programmes to satisfy compliance, not to build capability. The goal was documentation: hours logged, boxes ticked, an inspection passed. Whether a teacher actually taught differently the following week sat outside the measurement entirely.

That is not a criticism of the teachers who attended. You showed up. You took notes. Many of you genuinely wanted to change your practice. The system simply gave you nothing to change it with once the workshop ended.

If you are planning your own development this year, our guide to CPD for teachers and what actually counts separates activities that build skill from those that only build paperwork.

Understanding the Ebbinghaus Forgetting Curve in Teacher Training

Why does the workshop model fail so reliably? Ankit pointed to one of the oldest findings in memory research: the Ebbinghaus forgetting curve.

In the 1880s, the psychologist Hermann Ebbinghaus measured how quickly people forget newly learned material when they do not revisit it. The curve drops steeply at first, then flattens. Most loss happens almost immediately.

Ankit applied this directly to teacher training. Passive content delivered in a single sitting, he explained, loses around 70 percent of its retention within 24 hours and around 90 percent within a week. A teacher who sits through a full day on formative assessment on Saturday may carry a tenth of it into class by the following Saturday.

The curve is not destiny, though. Memory research consistently shows what flattens it: spacing learning over time, retrieving it actively, and applying it in real conditions. The workshop model does none of these.

Here is the uncomfortable irony. Teachers spend their working lives managing this exact curve for students. You build revision cycles, spaced quizzes, and recap starters because you know a single exposure does not stick. Then the profession trains you in a way it would never train a child.

Deliberate Pedagogical Practice: Lessons from the Medical Model

If forgetting explains why the old model fails, what explains growth? Ankit drew on more than a decade of Suraasa's teacher development data to answer that.

The finding cuts against common assumptions. Credentials alone did not predict which teachers improved. Years of experience did not predict it either. A teacher with fifteen years in the classroom can plateau in year three and simply repeat that year twelve more times.

What did predict growth was how a teacher treated their own pedagogy. The teachers who improved measurably approached teaching the way a good doctor approaches clinical practice. They tested a strategy. They studied what happened. They refined it. Then they tested again.

Medicine does not let a surgeon learn from a lecture and a certificate alone. Residency exists because clinical judgment only forms through supervised, repeated, reflective practice. Ankit's argument is that teaching deserves the same seriousness, and the data supports it.

A worked example. Imagine a Grade 8 science teacher who notices that the same five students answer every question. She decides to run an experiment for three weeks:

• Week 1: She keeps her usual hands-up questioning and tallies who responds. Twelve of thirty-two students speak at least once.
• Week 2: She switches to think-pair-share before every key question. Twenty-one students speak.
• Week 3: She adds no-hands cold calling after the pair discussion. Twenty-eight students speak, and the quality of answers rises because everyone has rehearsed.

No workshop gave her that result. Evidence from her own room did. That is deliberate practice: small, testable changes, tracked honestly. Our piece on bridging professional development and real classroom impact goes deeper into building this reflective loop.

Inside the Mind of the Modern Learner: 90-Second Attention Cycles

Ankit then turned to the other side of the desk. Students have changed, and he argued the change is physiological, not moral.

Around 2010, he noted, a student could sustain linear attention for roughly 25 minutes. Today's students operate differently. Their nervous systems, conditioned by years of short-form video and endless feeds, run in cycles of about 90 seconds.

This is the line from the webinar that teachers will recognise most:

"When you put a 40-minute lecture in front of them, you are not fighting laziness. You are fighting a trained nervous system." (Ankit Khanna)

That reframing matters. If you read disengagement as laziness, you respond with discipline. If you read it as conditioning, you respond with design.

Design does not mean entertainment. It means building lessons in modules that respect how attention now works, then using each reset as a chance to pull students deeper rather than lose them.

Notice what this structure does. It does not shorten the thinking. It segments the attention. The total time on a single hard idea can still run fifteen minutes or more, but no single stretch asks a student to sit passively beyond their cycle.

Parallel Intake vs. Deep Single-Threaded Processing

Ankit was careful not to frame today's learners as deficient. They have real strengths. They move fluently across multiple tabs, feeds, and sources at once. They scan quickly. They pull information from several places and assemble it.

Call it parallel intake. It is a genuine skill, and in many workplaces a useful one.

The cost shows up elsewhere. Students who excel at parallel intake often struggle with deep, single-threaded processing: holding one line of reasoning in mind and following it step by step to the end. Ankit gave two examples that every subject teacher will know. Following a complex mathematical derivation. Tracking an extended argument through a long text.

Both demand that the mind stay on one thread, carry earlier steps forward, and resist the pull to jump elsewhere. That capacity weakens without exercise.

What this looks like in a maths room. A teacher presents the derivation of the quadratic formula on the board, line by line. Students copy it. Most can reproduce it in a test. Few can explain why completing the square works, and fewer can adapt the method to a new problem.

Now flip it. The teacher gives students only the first and last lines, plus one hint: "What would make the left side a perfect square?" Students work in pairs to build the middle. Some stall at step two. The teacher circulates and asks questions instead of giving answers. Twenty minutes later, every pair has a derivation, and every student has had to hold the thread themselves.

The second lesson is slower. It is also the only one that trains single-threaded reasoning.

Cognitive Offloading: How AI Weakens Problem-Solving Resilience

This is where AI enters Ankit's argument, and where it becomes most urgent.

The human brain is an energy-conserving organ. Given an easier route, it takes it. That served our ancestors well. In a world where an AI tool answers any question instantly, it creates a problem.

When students hand off factual recall and quick problem-solving to AI, they remove the cognitive friction that builds critical thinking and spatial reasoning. Ankit opened the webinar with an analogy that captures it:

"AI is GPS for thinking. The thing that is withering away is not just memory or facts. It's the ability to struggle through a problem." (Ankit Khanna)

Think about what happened to wayfinding. Many people who rely on navigation apps can no longer form a mental map of their own city. The app did not make them worse at driving. It made them unpractised at orientation, because they never had to orient.

Students using AI to answer homework questions face the same risk. The answer arrives. The struggle that would have built the capacity never happens.

Global bodies are taking this seriously. UNESCO's work on artificial intelligence in education stresses a human-centred approach that protects learner agency, and the U.S. Department of Education has published guidance urging that AI keep teachers and human judgment at the centre of learning.

None of this means banning the tools. It means deciding, deliberately, where the friction stays.

The Death of the Explainer: Why Information Delivery Is Obsolete

Put the pieces together and Ankit's conclusion follows. If a student can get a tailored explanation of any concept, at their level, in their language, in seconds, then the teacher whose main value was explanation has lost their edge.

He described the old model bluntly: moving information from the textbook to the notebook. For generations that was a legitimate job, because access to clear explanation was scarce. Today it is abundant.

This is hard to hear, especially for teachers who are brilliant explainers. Explanation is a real craft. But being honest about where value now sits is a form of respect. You deserve to invest your energy where it matters most, not where the machine already wins.

Explanation still has a place. A well-timed, ninety-second clarification after students have wrestled with a problem lands far better than the same explanation delivered cold. The shift is in sequence and purpose. Explanation stops being the lesson and becomes one tool within it.

For a wider view of how AI is reshaping the role, see our guide to AI in education and what teachers should do about it.

So if explanation is no longer the core job, what is?

Engineering Productive Struggle in the Modern Classroom

Ankit's answer: the teacher's real work is now to place students in structured confusion and keep them there long enough to think deeply.

"Making it easy is the worst thing that a teacher can do, because the student is already doing that. He has every tool to make it easy. What they don't have is a structured difficulty that builds the mind." (Ankit Khanna)

The key word is structured. Productive struggle is not throwing students into the deep end and walking away. Unstructured difficulty produces frustration and shutdown. Engineered difficulty produces learning.

Here is a practical protocol you can adapt to almost any subject:

1. Pick the right problem. It should sit just beyond what students can do alone, but within reach with effort. Too easy and there is no struggle. Too hard and there is no progress.
2. Set a device-free window. Before anyone searches or asks an AI tool, students get a fixed period, say five minutes, to attempt the problem on their own.
3. Make first attempts visible. Mini whiteboards or sticky notes let you see the range of thinking at a glance.
4. Probe, do not rescue. When a student is stuck, ask a question that narrows the path: "What do you know for certain?" or "What did you try first?"
5. Add a second, harder variant. Once most students crack the first problem, extend it so the struggle continues on firmer ground.
6. Only then, bring in tools. Let students check their reasoning against an AI explanation and critique where it agrees or differs. Now the tool sharpens thinking rather than replacing it.

A worked example from a history class. Instead of asking "What caused the First World War?", which any chatbot answers in a paragraph, a teacher hands out three primary sources that point to different causes. Students must rank the causes and defend their ranking using only those sources. Only after the debate do they compare their reasoning with an AI summary. Most quickly spot what the summary flattens or misses.

This approach maps directly onto the higher levels of cognition. Our guide to Bloom's Taxonomy in the classroom shows how to write objectives that demand analysis and evaluation rather than recall.

The Three Teacher Archetypes in the Era of Generative AI

How are teachers responding to all this? Ankit described three broad patterns.

The Resister is often acting from integrity. They see the risks clearly. But refusing the tools entirely means spending hours on tasks that could take minutes, and losing touch with the environment students actually live in.

The Outsourcer is the mirror image. They adopt every tool and let it think for them. Lesson plans, worksheets, and report comments all come straight from a prompt. The output looks professional. But a teacher who outsources judgment has nothing left that a machine cannot offer.

The Augmented Teacher sits between them, and not as a compromise. It is a distinct and deliberate stance.

Defining the Augmented Teacher: Skills Machines Cannot Replicate

Ankit described the Augmented Teacher as someone who has done the hard thinking about where they add value and where they do not. In his words, this teacher has asked one essential question: what do I do that AI cannot do?

The answer clusters around a few human capabilities.

• Reading the room. Noticing the student who has gone quiet, the slump in energy after break, the half-second hesitation that signals a misconception. No dashboard catches these cues in real time.
• Building relationships. Students take intellectual risks for teachers they trust. Productive struggle only works if a student believes being wrong is safe.
• Making nuanced judgment calls. Knowing when to push and when to pause, when a struggling student needs a hint and when they need two more minutes.
• Designing difficulty. Choosing the problem that sits at the edge of a class's ability, and sequencing it so the struggle stays productive.
• Diagnosing conceptual bottlenecks. Spotting the root misunderstanding underneath a wrong answer, which digital diagnostic tools often miss.

Meanwhile, the Augmented Teacher uses AI without hesitation for the administrative layer: first drafts of lesson outlines, rubric design, differentiated versions of a worksheet, parent email drafts. Every minute saved there becomes a minute available for one-on-one intervention in class.

Practical starting points are in our guides to AI tools for educator productivity and ChatGPT for lesson planning, assessment, and productivity.

The rule is simple. AI drafts. You decide.

The Three-Pillar Framework for Lifelong Teacher Growth

If the old model of development fails, what replaces it? Ankit closed with a framework built on the same medical-residency logic he used earlier. Teacher mastery, he argued, rests on three pillars.

Pillar 1: Sustained practice over months

Real change in practice takes time. A skill like questioning for deep thinking, or designing productive struggle, forms over many lessons, not one afternoon. Development needs to run long enough to beat the forgetting curve and turn new strategies into habits.

Pillar 2: Iterative feedback on recorded teaching

Recording your own lessons is one of the most powerful and least used tools in the profession. You cannot fully observe yourself while teaching. Watching a recording reveals the questions you rushed, the students you overlooked, and the moments where struggle tipped into confusion.

Try this: record one lesson a fortnight. Watch ten minutes of it with a single focus, such as wait time after questions. Change one thing. Record again. That is clinical practice applied to pedagogy.

Pillar 3: Recognised credentials that validate demonstrated competence

Credentials matter, but only when they certify what a teacher can actually do in a classroom. Ankit stressed credentials that validate demonstrated pedagogical competence and carry recognition globally, so the skill you build in one school travels with you to the next.

This is the thinking behind our Professional Graduate Certificate in Teaching and Learning (PgCTL): long-form, practice-led development rather than a one-off course. If you are comparing options, our breakdown of the best teaching certifications for career growth sets out what to look for.

Alongside credentials, build a portfolio of evidence: lesson plans, recordings, student work samples, and reflections showing how your practice changed. International schools increasingly want proof of impact, not just qualifications on paper.

Why AI Will Evolve the Profession Rather Than Replace Teachers

It is easy to read headlines and assume the worst. Ankit's argument points the other way. AI is not removing the need for teachers. It is removing the parts of teaching that were never the heart of it.

What remains is harder, more human, and more valuable. Designing difficulty. Reading a room. Holding a student in productive confusion and knowing exactly when to help. Building the trust that makes learning possible. International research, including the OECD's Teaching and Learning International Survey, consistently shows that the quality of teaching practice and professional collaboration shape student outcomes. Those are human variables.

Here is how to start this week:

• Audit one lesson and break any passive stretch longer than a few minutes into modular segments.
• Add a five-minute device-free struggle window before students reach for search or AI.
• Use AI to draft one rubric or differentiated worksheet, then refine it with your own judgment.
• Record one lesson and watch ten minutes with a single focus.
• Start a simple portfolio folder for evidence of changed practice.

You entered this profession to shape minds. The tools have changed. That purpose has not. If anything, the future of teaching skills asks more of you as a professional, and rewards you more for meeting it.

You do not have to figure it out alone. We built Suraasa as the system that starts with the teacher, and stays with them.

For the Love of Teaching.