AI and Education - Part 1, The New Reality

🤖 The New Reality

AI is here, the genie is out of the bottle, and it won’t be going back. The task now is to harness it to support learning and make it an integral part of what we do.

Education has always adopted new technologies, but the wide-scale use of AI is proving more disruptive than most expected. Promises of a new “utopian” system—where everyone has their own personal tutor and learning journey—are unlikely to be fully realised.

History shows us that new technologies often amplify existing inequalities. AI in education carries the same risk, even as it promises greater equity.

Motivation for education is always something that underpins equity in education. Currently the equity and motivation aspects of education are taking a back seat, to being able to focus on those that already want help.

🤔 The current state

Generative AI currently dominates education. Students use it to solve problems, write assignments, and prepare study notes. Educators use it to generate lessons, exams, and support material. But this is still early-stage adoption and often its issues come to the surface quickly.

Silicon Valley is now funnelling billions into specialised applications built on top of Large Language Models (LLMs) like those from OpenAI, Google, and Meta. These include:

• Tutors – apps that answer student questions.
• Marking – systems that grade essays and responses.
• Content generation – tools that produce quizzes, flashcards, and summaries.
• Multimedia – apps that convert videos, images, or audio into structured notes.

The key issue: are these tools genuinely supporting learning, or just helping complete tasks?

Over the coming articles we'll explore these applications and their impact on education.

This is the unresolved challenge.

🙋 The smart, available tutor

The first application we’ll explore is the tutor. A tutor’s role is to assist in learning, but most major AI tools (ChatGPT, Co-Pilot, Gemini) won’t tutor; they’ll answer. The primary mechanism of an LLM is to answer, not to tutor.

Applications built on top of these systems can harness the power of LLMs to transition from answering to tutoring. To date, we’ve seen the ability of AI to perform at the highest level in the Maths Olympiad and other standardised exams.

LLMs are being shown to be able to answer questions to a high level of accuracy, even with complex arrangements.

Answering questions is a task at which AIs are excelling. They will still make mistakes, and as seen in the above example, it is only the highest-level AIs that are able to perform at human levels in exams like the Maths Olympiad.

Let’s look at a few examples of how the LLMs respond to a tutoring question:

Let’s ask the question,

How do you solve the quadratic equation $x^2+5x+6=0$?

Open AI - Chat GPT 5

You can solve

$$x^2+5x+6=0$$

by factorising: Look for two numbers that multiply to 6 and add to 5. ✅ $$2×3=6$$ $$2+3=5$$

Rewrite: $$x^2+5x+6=(x+2)(x+3)$$ Set each factor equal to zero:

$$x + 2 = 0 \implies x = -2$$

$$x + 3 = 0 \implies x = -3$$

✅ Final Answer: $x=−2$ or $x=−3$

Do you want me also to show you how to solve it using the quadratic formula as a check?

---

Google - Gemini 2.5 Pro

The solutions to the equation $x^2+5x+6=0$ are $x = -2$ and $x = -3$. There are a couple of common ways to solve this type of equation: factoring and using the quadratic formula.

Solving by Factoring

Find two numbers that multiply to 6 (the last term) and add up to 5 (the middle term). In this case, the numbers are 2 and 3 (since 2times3=6 and 2+3=5).

Rewrite the equation using these numbers. The equation becomes (x+2)(x+3)=0.

Solve for x. For the entire expression to be zero, one of the parts in the parentheses must be zero.

If $x+2=0$, then $x = -2$.

If $x+3=0$, then $x = -3$.

Solving with the Quadratic Formula

The quadratic formula can be used to solve any quadratic equation in the form ax2+bx+c=0.

The formula is: $$x = \frac{-b \pm \sqrt{b^2 - 4ac}}{2a}$$

Identify a, b, and c in your equation. In $x^2+5x+6=0$, we have a = 1, b = 5, and c = 6.

Plug these values into the formula. $x = \frac{-5 \pm \sqrt{(-5 )^2 - 4(1)(6)}}{(2)(1)}$

Simplify the equation. $x = \frac{-5 \pm \sqrt{25 - 24}}{2}$

Find the two possible solutions.

$x=−2$

$x=−3$

---

Intuition - Intu AI - 2.5t (Stella)

Alright, let's tackle this quadratic equation! Solving an equation like means we're looking for the values of that make the statement true.

There are a few key methods from our HSC Advanced Maths syllabus to solve quadratic equations:

• Factorisation: If the quadratic expression can be factored easily.

• Quadratic Formula: A universal method that works for any quadratic equation of the form .

• Completing the Square: Another general method that can transform the quadratic into a more solvable form.

To begin, which of these methods do you think would be the most suitable or easiest to apply for this particular equation? 🤔

User Response: "Factorisation"

Excellent choice! Factorisation is often the quickest method when it's applicable, and it's a fundamental skill in Algebra.

When we factorise a quadratic expression like $x^2 + bx + c$, we're looking to express it as a product of two linear factors, like $(x+p)(x+q)$

The key idea here is that for this to be true, the numbers and must satisfy two conditions:

• Their product $p \times q$, must equal the constant term $ c $
• Their sum $p + q$, must equal the coefficient of the $x$ term, $b$.

For our equation $x^2+5x+6=0$,what two numbers do you think multiply to give $6$ and add to give $5$? ✨

---

Answering vs. Teaching

The LLMs are adept at answering, and there appears to be an attempt at teaching and guiding. The Intu AI example attempts to simulate the tutoring and Socratic experience, involving asking and challenging questions.

Intu AI - ASK is our implementation of using an LLM, as well as our own content (which augments the knowledge of the AI), to tutor students. One key part of our AI is the introduction of the tutor, Stella.

🤵 Humanising the AI

Our first version of Intu AI was essentially a wrapper around LLMs. It worked—but it felt impersonal, like using a really good Google.

The breakthrough came when we built tutors with distinct personalities, inspired by our real staff. Rapport and choice matter in education. Students want a tutor who “gets” them and knows how to work with them.

The ability to try and then choose your tutor is an integral part of education and the personalised education journey. We introduced three tutors, each with their own personality, based on characteristics of our tutors, but all grounded in the Intuition teaching methods.

• Stella – Guiding but challenging; ensures deep understanding.
• Cassie – Concise and direct; efficient with clear answers.
• Ella – Supportive and encouraging; focuses on confidence.

Students should find a tutor they prefer and stick with them. The graph below shows the breakdown of tutor choice (note that Stella is the default tutor)

The breakdown of choices students make for their tutor (as of August 2025)

Giving students choice builds engagement. However, another challenge remains: students often treat AI tutors with a sense of detachment, as if they were bots. Our data shows that when they use conversational, respectful dialogue, outcomes improve.

However, we still have work to do. The next phase of this engagement is to move students away from treating the AI as a bot and more towards treating it as a person.

The blunt and at times rude conversations frame the learning. When there is little respect between the two, a lack of rapport already exists, which we know is vital for learning.

The humanisation of AI will transition it from being a 'throwaway' tool to a trusted mentor. However, this transition requires additional work and research.

---

Building a connection

Sci-fi shows us that the line between human and AI will be blurred, and it already is for many people. As we move towards the humanisation of AI responses, we need to be aware of the social and mental impacts.

Do we want to have people who are blunt and demanding of those who are trying to help? Well, that's what we're conditioning students to do. Being polite and respectful might seem like a waste when it's just a machine, but it forms the basis of our relationships and interactions.

So, where to from here? We envision the future of tutors as more human and deserving of respect. When the relationship is based on rapport, respect, and safety, the outcomes for students will improve.

The future of AI tutoring lies in deeper, more human-like connections—through voice, avatars, and even video. Education has always thrived on human connection, from Socrates onwards. AI tutors are the next step in that tradition.

🕵️ What's next?

In the next part, we’ll look at the generation of personalised learning with generative AI. In the coming articles, we’ll also explore human-in-the-loop education and the new way to assess.