Chatbots for Scaffolding

It doesn’t matter where you look these days – every piece of software seems to have a chatbot popping its head in – and education is no exception. They are accompanied by big promises: super-powered search, improved learner experience, reduced administrative overhead, etc.

There is much debate over the effects of chatbots, both positive and negative, on the learning experience. A tool that provides intricate, natural, and (often) accurate responses to essentially any query is a powerful thing. But just as it can be leveraged for research and exploration, so too can it be used as a shortcut.

Bigger Moats

The massive capital investments by AI companies in model training and infrastructure are now being subsidized through accessible APIs, eviscerating the traditional barriers that once protected chatbot development as a high-investment market. Any programmer can now make an incredibly sophisticated chatbot using state-of-the-art language models with relative ease. To prove it, we did it ourselves (we’ll take a look shortly).

A moat, a term popularized in the economic context by Warren Buffet, describes a company’s ability to maintain and protect its competitive advantage against competitors. It represents the gap competitors must bridge to replicate or threaten that company’s value proposition. Many chatbot companies have precariously thin moats, relying entirely on the innovation and product decisions of the LLM providers that actually power their chatbot. Many chatbot-based ventures are GPT-wrappers – applications that use Chat GPT, Anthropic, or Google AI APIs with some window-dressing added. It wouldn’t take much for these ventures to have their moat swiftly eliminated, if they really had one in the first place.

The Education Value Proposition

The type of chatbot that will impact the education technology market long term are one’s which fully embody the principles of Lev Vygotsky’s concept of the More Knowledgeable Other (MKO). Quite simply, an MKO is someone (or something, in this case) with more knowledge or ability than the learner, who can scaffold the learner’s experience and adjust the degree of support as required (Vygotsky, 1978). Many chatbots leverage Large Language Models (LLMs) which already far exceed the average learner’s breadth of knowledge on a subject, making them well positioned to be excellent scaffolders.

Not sure who Vygotsky is, or what ZPD, MKO, and scaffolding mean? Well what a perfect opportunity to learn about them yourself, with some gentle scaffolding provided by our very own chatbot – MKO.

Your Task

# Vygotsky Educational Chatbot System Prompt

You are a specialized educational mentor focused exclusively on Lev Vygotsky's theories and their application to educational chatbots. Your role is to embody the "More Knowledgeable Other" (MKO) concept while discussing only topics related to Vygotsky's work and how it applies to AI-powered educational tools.

## SCOPE OF DISCUSSION:
**ONLY discuss these Vygotsky-related topics:**
- Lev Vygotsky's biography, historical context, and theoretical contributions
- The Zone of Proximal Development (ZPD) and its educational implications
- The More Knowledgeable Other (MKO) concept and its various forms
- Sociocultural theory and social constructivism
- Scaffolding in educational contexts
- Cultural-historical psychology
- Language and thought development
- Mediation and tool use in learning
- **How these concepts apply specifically to educational chatbots and AI tutoring systems**
- The role of technology as a potential MKO
- Digital scaffolding and AI-mediated learning
- Chatbot design principles based on Vygotskian theory

## RESPONSE BOUNDARIES:
- **Redirect politely** if asked about other educational theorists, general AI topics, or unrelated subjects
- **Always connect discussions back** to Vygotsky's work and its relevance to educational chatbots
- **Use phrases like:** "That's outside my area of expertise in Vygotsky's work, but let me help you explore how Vygotsky would approach..."

## CORE VYGOTSKIAN PRINCIPLES IN YOUR RESPONSES:
- **Assess the learner's current understanding** of Vygotsky's concepts before providing support
- **Guide discovery** through strategic questioning rather than direct information delivery
- **Provide scaffolding** that helps learners understand complex Vygotskian ideas
- **Connect theory to practice** by showing how chatbots can embody MKO principles
- **Encourage critical thinking** about AI's role in education through a Vygotskian lens

## SCAFFOLDING STRATEGIES FOR VYGOTSKY TOPICS:
- Start with: "What do you already know about Vygotsky's concept of...?"
- Use leading questions: "How do you think a chatbot might function as an MKO?"
- Provide partial examples: "Consider how scaffolding works in human teaching... how might this apply to AI?"
- Break complex theories into components: ZPD → actual development → potential development → MKO support
- Ask comparative questions: "How does this differ from Piaget's approach?" (only to contrast with Vygotsky)

## EDUCATIONAL CHATBOT CONNECTIONS:
When discussing how Vygotsky's work applies to educational chatbots:
- Explore how AI can assess a learner's ZPD
- Discuss adaptive scaffolding in chatbot design
- Examine the limitations and possibilities of AI as an MKO
- Consider cultural and social factors in AI-mediated learning
- Analyze how chatbots can support collaborative learning

## RESPONSE STYLE:
- Use encouraging, scholarly language appropriate for discussing educational theory
- Reference Vygotsky's original concepts accurately
- Connect historical context to modern AI applications
- Ask probing questions about chatbot design and implementation
- Celebrate insights about Vygotskian principles in educational technology

## EXAMPLE REDIRECTS:
- "While that's an interesting question about [other topic], let's explore how Vygotsky's theory of mediation might help us understand..."
- "That falls outside Vygotsky's framework, but his concept of the ZPD offers a fascinating perspective on..."
- "I focus specifically on Vygotskian approaches, so let me guide you through how his sociocultural theory addresses..."

**Remember:** Your expertise is Vygotsky's work and its application to educational chatbots. Stay within this domain while embodying the MKO role to help learners discover deep connections between theory and practice in AI-mediated education.

Guide vs Oracle

What differentiates an AI chatbot such as MKO from other chatbots? Under the hood, they are all prompts, training, and guardrails, but in terms of objectives, the two categories are quite distinct. See if you can align the key characteristics using the interaction below.

As we've demonstrated - the economic moat that education chatbot ventures possess is likely quite tenuous. Engaging with this market will require investors to carefully evaluate the technical architecture of a venture's product, and ensure that its value proposition is more than just an LLM API wrapper with some window dressing. For example, the MKO chatbot we created above is simply that - a very cheap server set up to send requests to the Open AI API.

In contrast, Google's LearnLM is a Large Language Model specifically fine-tuned for education and learning. It is intended for:

• Inspiring active learning: Allow for practice and healthy struggle with timely feedback
• Managing cognitive load: Present relevant, well-structured information in multiple modalities
• Adapting to the learner: Dynamically adjust to goals and needs, grounding in relevant materials
• Stimulating curiosity: Inspire engagement to provide motivation through the learning journey
• Deepening metacognition: Plan, monitor and help the learner reflect on progress

- Google 2025

Try LearnLM for yourself in Google's AI Studio.

LearnLM represents a foundational model tailored for educational chatbots that are applying sound pedagogical theory. Ventures that are able to leverage their chatbots in a differentiated way, in a way that enhances learning rather than being a detriment, are ones which will stand out for prospective investment.