How AI Will Revolutionize Learning and Education 🧑‍🎓

Guest Post in Impact Supporters #3

Intro from August Solliv 👋

Once a month, I invite a friend from the impact VC landscape to write a guest post for Impact Supporters on a specific angle in the impact VC space. These articles are a bit more niche and specific on certain topics. Hope you find it interesting!
This time, I invited Kim to write about AI in edtech. As you might remember from my article with Jonathan from Educapital, this is one of the key areas in edtech - find the article with Jonathan at the end of the article 🤗. Kim is a super promising founder in this field building Alice.tech. Really hope you enjoy it 💥

If you want to be the next person writing a guest post in Impact Supporters, feel free to text me with your idea and we can discuss the opportunity! E-mail is august.solliv@gmail.com.

Intro about guest post author 👋:

Kim Rants is CEO and Co-founder of Alice.tech – an innovative AI-startup that makes learning effective and simple through personalization. He was previously Associate Partner in McKinsey, where he specialized in AI, has been Head of Global Business Development at LEGO, and has been teaching university students for >11 years.

Setting the scene 🖼️

Today's education model is outdated. Born in the Industrial Revolution, the "assembly-line" model of schooling was designed for uniformity. But in a world of rapid technological change, why are we still stuck with a one-size-fits-all approach when every learner is unique? 🤔

Until recently, there was a simple reason: solving this problem was technically and economically impossible. But with the latest advancements in AI, personalized high-quality learning for all is now within reach. This shift has the potential to revolutionize education—improving equality, enlightenment, and positive global change. 🌍

As someone who has experienced education as both a student and a teacher, and who is now CEO and Co-founder of an exciting EdTech Startup (Alice.tech), I’m deeply excited about this revolution. In the sections that follow, I’ll dive into (i) the current challenges in education, (ii) why now is the turning point, (iii) why AI’s impact will be positive if we do it right, (iv) how we can ensure this revolution is responsible, and (v) my tips for you as an impact investor in AI-driven EdTech.

Have fun reading! 🙏

(i) The Current Challenges in Education

Education is one of the most critical and largest industries globally, with expenditures expected to reach around $10 trillion by 2030. The current system works “well enough,” delivering incremental progress; every year, society gets a bit smarter, and more people are lifted out of poverty. That’s undeniably positive. 👏

However, this model has one fundamental flaw: lack of personalized learning. Learners are all unique (different motivations, skillset, starting points, learning styles, etc.). It’s difficult to estimate the exact potential of personalized learning, but it is impossible to deny that the potential is tremendous.

As a teacher, you might have 20, 30, or sometimes 100+ students to cater for. This means that teachers need to tailor their lecturing to some artificial “average” across the students (difficulty, pace, etc.). This results in the poorest performing students being dropped behind, and the best students not getting challenged enough.

When individual students try to correct for this, they currently have different options available to them, such as human tutors or online resources.

Human tutors are generally quite expensive (in most countries only available for upper middle class and up) and of very varying quality. Little innovation has happened in this area other than some companies being successful in establishing scale and having brands where a certain quality is promised.

For online resources, some progress was made in the 2010s with the digitalization of education, particularly in making learning resources more accessible. However, much of this digital content is still standardized and disconnected from the specific needs of individual learners.

Institutions, which control the credentials and official recognition that students need, continue to dictate much of the learning journey—and that’s not necessarily a bad thing. But it means that standardized digital content, while useful, often becomes an “extra” resource rather than a tailored solution. For many students, this can be overwhelming. As an example, they don’t necessarily need a YouTube lecture from a brilliant professor explaining a similar topic in a different way. What they really need is personalized help that aligns with exactly what they’re learning in their course—something that adapts to their level, fits their schedule, and caters to their unique learning style.

This is where AI has the potential to revolutionize learning. Instead of static, pre-created content that quickly becomes outdated, EdTech companies in this new paradigm can focus on developing robust AI engines that dynamically adapt to each learner’s context. This personalization can be scalable and cost-effective, opening the door to individualized learning. 🚀

(ii) Why now is the turning point

Just two years ago, building a centralized AI model capable of delivering personalized learning across diverse study contexts seemed out of reach. But today, with the rapid advancement of large language models (LLMs) and the rapid adoption, I’m convinced that AI can revolutionize learning and education.

The pace of AI improvement is exponential, which covers both quality and cost. Some examples include:

• Costs of AI tokens are plummeting ~80% per year (https://x.com/AndrewYNg/status/1829190549842321758)

• OpenAI reported that GPT-3.5 ranked in the 31st percentile compared to high school students competing in the Biology Olympiad, whereas GPT-4 scored significantly better, scoring in the 99th percentile

Coupled with this, we are seeing explosive adoption of LLMs. It appears that GenAI has made AI much more accessible to people – all of the sudden it is not just computer scientists that can reason with the output; the models themselves portray as approachable creatures that are easily engaged with by the layman.

We have all seen how ChatGPT broke all records in getting to 100mn users in just 2 months, but as an anecdotal datapoint, we recently surveyed 200+ students and found that 100% had tried AI for their studies. This kind of widespread adoption underscores that AI is no longer a future technology—it’s a tool that students are already turning to for help.

This confluence of factors—exponentially improving technology, rapidly falling costs, and growing adoption—marks a critical turning point for AI in education.

(iii) Why AI’s impact will be positive if we do it right

AI brings unprecedented personalization, scalability, and democratization to education. If we do it right, it can benefit all stakeholders in the ecosystem—students, teachers, institutions, and publishers alike.

For students, AI will make effective learning simpler than ever. AI tailors learning to each student's pace, interests, and skill level, creating engaging and adaptive experiences that are truly personalized. 🎯

For teachers, AI can be the TA they always wanted. It can help prepare lessons, create exercises, and give real-time insights into class performance, pinpointing where extra attention is needed. This frees teachers to focus on what matters most: mentoring students who need additional support and setting the overall direction. Importantly, AI shouldn’t replace human teachers—it enhances them. The goal isn’t to automate learning, but to empower teachers with tools that make them more effective. AI can reveal hidden patterns in student behavior and performance, helping teachers better understand students’ strengths and weaknesses, even for those who might not speak up during classes. 🧑‍🏫

For institutions, AI can streamline operations and reduce administrative burdens (e.g., automating grading while removing potential biases), allowing institutions to reallocate resources to where they can make the biggest difference for student learning and wellbeing 🏛️

For publishers, AI creates new opportunities for engagement. It can help turn static educational materials into dynamic, interactive experiences tailored to individual learners, increasing the value and relevance to students. And, importantly, AI can be built in a way where it shouldn’t cause losses to the publishers – for example by not training on the input material and not sharing output among students, thereby potentially causing copyright problems (as many free models out there do) 🏢

(iv) How we can ensure this revolution is responsible

With all the promise that AI holds, it’s crucial that we deploy it responsibly. AI must enhance learning while upholding ethical standards. 📜 Here are some key principles I believe should guide us (not exhaustive):

• Combating Plagiarism: AI should foster genuine learning, not be a tool for shortcuts or cheating. Systems must be designed to encourage critical thinking and originality. BUT(!), as society, we also need to let go of the romantic thought that students always need to read everything to get a good grade. Let’s measure student output and not input. 👆

• Bias Awareness: While AI can reduce human biases, it’s not immune to them. Ensuring fairness must be at the core of every system we build (e.g., for grading).

• Human in the Loop: Teachers remain irreplaceable. AI should augment their capabilities, not take over the entire learning process. The human connection between teachers and students is essential.

• Test and learn: This is potentially the biggest one. But I think it is critical that we don’t believe we can come up with the perfect solution upfront – that’s impossible and will only “paralyze” us. Then nothing will happen. We need to start in the small (e.g., one class out of a full school) and then gradually roll out AI in a controlled manner. Let’s iterate and learn. 🛠️

Achieving responsible AI integration requires collective action from all stakeholders:

• Startups and Investors: Companies (like ours) need to focus on building platforms that prioritize deep, meaningful learning rather than "quick-fix" solutions. While some AI tools have gone viral by offering shortcuts, real learning is never effortless. AI can make learning more efficient and engaging, but not effortless. We must resist the temptation to build tools that promise results without effort.

• Regulators, Publishers, and Institutions: While it’s important to regulate AI, we must avoid stifling innovation. AI is here to stay, and its potential in education is immense. Too much caution could lead to missed opportunities, so regulation should balance safety with innovation. My hope is that publishers and institutions will seek collaborations rather than creating walled gardens to try and block out innovation (a fight they are destined to lose over time, I believe. Students always find a way to the best technology).

• Teachers: Educators should embrace AI by starting with simple applications like performance insights and gradually incorporating more advanced use cases, such as lesson planning assistance. AI can be a powerful ally already in making their work more efficient and impactful.

• Students: Most students are already using AI, and they should continue to do so—responsibly. AI shouldn’t be seen as an easy way out for assignments or tests. Instead, it should be a practice companion and critical thinking partner, helping identify weaknesses and providing targeted support. The students who learn to use AI as a learning aid, rather than a shortcut, will be the high achievers of the future.

(v) My tips for you as an impact investor in AI-driven EdTech

So, should you invest in EdTech? 🤔

Based on the points above, I hope you’re seeing the immense potential. While EdTech is a promising field, let’s be honest about the challenges:

• Students have limited budgets, are seasonal users (difficult to retain during holidays), and naturally churn when they finish school. While “lifelong learning” is an opportunity, it’s still a hurdle.

• Institutions are slow-moving and often unable to pay substantial fees.

• Teachers may be less inclined to adopt new technologies compared to their students.

However, a major challenge has been overcome: scalability. Previously, scaling in EdTech was tough as content quickly became outdated and needed a super high degree of localization. With AI, though, it’s now possible to build a “central AI engine” that scales across borders segments.

That’s why I believe now is the right time to invest in EdTech—you could contribute to creating a unicorn that also does good for the world.

But, are AI-driven EdTechs not just “AI wrappers?” 😱

Maybe. But does it matter? A year ago, “AI wrapper” was a term AI-skeptical investors used frequently, but the conversation has luckily become more nuanced. The key to a successful AI application—or “wrapper”—is going deep within a domain, and Sequoia’s framework below highlights this well.

If you just build a generic chat bot or a summarizer (which I have already observed many times), chances are you will be steamrolled by OpenAI & friends.

Source: Sequoia, 2024

However, if you go deep within a domain you stand a chance. While LLMs are incredibly powerful, they remain general-purpose tools. In a specialized domain like learning and education, a general AI model will never be the optimal solution.

In EdTech, this depth translates to:

1. Address core student pain points – Focus on the real challenges students face.

2. Leverage LLMs effectively – If an advanced LLM can solve a problem, use it. But if a specialized approach is needed, that’s an opportunity for innovation.

3. Create value beyond AI – True EdTech success requires more than just AI. It’s about delivering an intuitive UX, building an emotionally resonant brand, and incorporating unique features –or full systems– like our “Learning Loop” at Alice.tech.

In short, AI-driven EdTechs will succeed by solving specific, high-value problems that complement LLMs rather than relying on them entirely. And this principle, in my view, goes beyond education alone.

So, should you go for B2C or B2B/B2G?
There’s no universal answer here, but starting with B2C and gradually shifting towards B2B might be the smartest path. Students are more likely to be early adopters, providing immediate feedback. By gaining traction with these users, EdTech companies can refine their products, build brand loyalty, and gather valuable feedback.

Institutions, especially traditional ones like universities, will likely take longer to fully adopt AI-driven solutions. By balancing B2C and B2B approaches, EdTech companies can achieve early growth with long-term, scalable impact.

Therefore, I recommend to look for a balanced approach: B2C provides early momentum and feedback, while B2B enables long-term, scalable impact that benefits all stakeholders. But hey, I might also be biased from how we see the world at Alice.tech.

Final thoughts

In conclusion, AI is far more than just a buzzword—it has the potential to improve the learning of all students, bring more equality, elevate teachers, and benefit other stakeholders in the learning ecosystem as well.

However, we must embrace its potential thoughtfully, ensuring that AI is rolled out responsibly. This starts with being aware of potential pitfalls (plagiarism, etc.) and then it is critical to do the roll-out gradually, where we allow us to test and learn.

This could be just the beginning of a major shift in how we learn—and as a founder in this space, I’m excited to be at the forefront of this transformation. Together, let’s create a future where learning is more personalized, accessible, and impactful than ever before.

Follow us at Alice.tech—our journey has only just begun. 🚀

Article with Jonathan from Educapital on edtech if you want to also get the VC perspective on edtech:

Thanks for reading this guest post! Hope you found it interesting and that you like these more in-depth articles with super knowledgable guests.
See you soon for the next weekly issue, and next month for the next guest post! 👋