Unlocking the Future of Writing with Structured Knowledge
Unlocking the Future of Writing with Structured Knowledge
How I’m preparing the next generation of writers
Today, we're diving deep into a topic that I’ve been considering for a while that is actively reshaping how I approach teaching professional writing: the power of structuring knowledge for smarter AI interactions.
I recently did a 10 minute presentation for BoodleBox and wanted to expand my thoughts here on Cyborgs Writing. You can see the short version in the recording of BoodleBox’s recent AI convocation. My presentation is at the 1 hour mark.
This isn't just about tweaking our prompts or fine-tuning our AI tools; it's about fundamentally rethinking how we prepare the next generation of writers for a world where AI is an integral part of the content creation process.
As many of you know, I teach Introduction to Professional Writing at the university level. For years, we've started our courses with the basics: "What makes a good sentence?" We've focused on crafting clear, concise prose as the foundation of effective communication. But the world of professional writing is evolving rapidly, and our teaching methods need to keep pace.
This semester, I made a significant change. Instead of starting with sentences, we kicked off our first week with lessons on structure. Why?
Because I believe that understanding and creating effective structures will be the core of writing in the future.
This shift isn't just about adapting to new technologies; it's about embracing a holistic view of content operations.
Embracing Content Operations
In revising my course, I've centered around the concept of content operations. This approach gets students to understand the entire lifecycle of content in their field, from ideation and creation to management, distribution, and analysis.
And yes, this lifecycle now includes AI and knowledge bases as crucial components … or will soon. It’s not a new idea, but has increased relevance in an AI world.
By focusing on structure from the outset, we're preparing students not just to write well, but to think systematically about how information is organized, accessed, and utilized across various platforms and technologies. This structural thinking is the foundation upon which effective AI interactions will be built.
By teaching students to think in terms of structure and content operations, we're equipping them with the skills they'll need to:
Craft more effective prompts for AI tools
Organize information in ways that AI can more easily process and utilize
Create and manage knowledge bases that can enhance AI performance
Understand the interplay between human expertise and AI capabilities in content creation
This focus on structure isn't entirely new. In fact, it harkens back to the principles of classical rhetoric, where orators carefully structured their speeches to guide audiences through complex ideas. What's new is how we're applying these timeless principles to cutting-edge technology.
Just as ancient orators shaped the world with words in forums and courts, today's professional writers shape digital realities through structured content and AI interactions. By teaching students to be architects of information, we're preparing them to be the world-builders and culture-shapers of the digital age.
Understanding the Language of AI through Vectors
To truly grasp why structuring our knowledge and prompts is so crucial for effective AI interactions, we need to dive into the concept of vectors.
Don't worry – we're not about to get lost in complex mathematics. Instead, let's explore how vectors form the foundation of how AI "thinks" and processes information.
In the context of AI and machine learning, vectors are essentially lists of numbers that represent pieces of information. Think of them as the AI's way of understanding and categorizing the world. Just as we might describe a car by its color, make, model, and year, AI represents concepts, words, or even entire documents as sequences of numbers.
When an AI model processes text (or any type of data), it converts that information into vectors. This process, often called "embedding," allows the AI to understand relationships between different pieces of information. Here's a simplified example:
The word "cat" might be represented as [0.2, -0.5, 0.8]
"Dog" might be [0.3, -0.4, 0.7]
"Automobile" might be [-0.8, 0.1, -0.3]
In this vastly simplified representation (these vectors become much more complex when trained on large corpora of text), you can see that "cat" and "dog" have similar vectors, reflecting their conceptual similarity as domestic animals. "Automobile," on the other hand, has a very different vector, indicating its distance from the animal concepts.
Now, you might be thinking, "That's interesting, but why should I care about vectors as a writer or educator?" Here's why:
Contextual Understanding: AI doesn't understand words the way we do. It understands relationships between vectors. When we structure our prompts and knowledge bases effectively, we're helping the AI create more accurate and useful vector representations.
Nuanced Communication: By understanding that AI processes information as vectors, we can craft our prompts to provide clearer context and relationships between concepts. This leads to more nuanced and accurate AI responses.
Efficient Knowledge Retrieval: Well-structured information allows AI to more efficiently navigate its vector space, leading to faster and more relevant retrieval of information.
Improved Pattern Recognition: Structured data helps AI models identify patterns more effectively, leading to better insights and more accurate predictions.
So, how do we apply this understanding to our interactions with AI? Here are a few strategies:
Use Clear Categories: When organizing information, use clear, distinct categories. This helps the AI create more differentiated vector representations.
Provide Context: Always give context in your prompts. Remember, the AI is looking at relationships between vectors, not just individual words.
Be Specific: The more specific and structured your input, the more precise the vector representations will be, leading to more accurate outputs.
Leverage Hierarchies: Use hierarchical structures in your knowledge bases. This mirrors how vector spaces often organize information, making it easier for AI to navigate and understand.
Consistent Formatting: Use consistent formatting and structure across your documents and prompts. This helps the AI create more coherent vector representations across different pieces of information.
By understanding vectors, we're not just using AI tools more effectively; we're learning to "speak" the language of AI. This understanding allows us to structure our knowledge and prompts in ways that align with how AI processes information, leading to more powerful, accurate, and nuanced interactions.
As we continue to integrate AI into our writing and teaching processes, this vector-centric approach to structuring information will become increasingly crucial. It's not just about adapting to AI; it's about shaping the very foundation of how AI understands and interacts with our content.
Structuring Knowledge in Action: A Classroom Example
To illustrate the power of structured knowledge and prompts, let's walk through an exercise I recently conducted with my students. This example not only demonstrates the importance of structure in professional writing but also shows how we can leverage this understanding to create more effective AI interactions.
I asked my students to find real examples of professional writing from their fields and examine the structure. Their task was to identify and label different parts of the content, essentially creating a template. I posed this question: "If we wanted to replicate this piece of writing, what would be the different parts?"
Let's look at a specific example: a social media post designed to help readers choose a statistical test for analyzing research.
Here's the structure we identified:
[HOOK] An attention-getter that highlights a need the audience has.
[ACTION] Something quick someone can do to fill that need.
[LIST] Usable list of items with short explanations of how they fulfill the need.
[METHOD] How to use that list to fulfill need.
[CALL TO ACTION] Engaging question to drive comments and expand the usefulness of this post.Now, let's see how this understanding of structure can improve our interactions with AI. Suppose we want to create a similar post about AI tools for people with ADHD.
First, let's try a simple, unstructured prompt:
Write a social media post about the top 5 AI tools for people with ADHD.
The result is okay, and the AI somewhat follows the structure because it has analyzed many similar posts. However, it's not quite hitting the mark.
Now, let's use the same prompt but include the structure we identified:
Write a social media post about the top 5 AI tools for people with ADHD. Use the following structure:
[HOOK] An attention-getter that highlights a need the audience has.
[ACTION] Something quick someone can do to fill that need.
[LIST] Usable list of 5 AI tools for people with ADHD.
[METHOD] How to use these tools effectively.
[CALL TO ACTION] Engaging question to drive comments.
The results are much better. The AI now has a clear framework to follow, resulting in a more focused and effective post.
But we can take this even further. What if we provide the AI with specific, curated information about AI tools for ADHD? We can give it a list of tools and descriptions that we've personally used and vetted:
<description> Rize is an AI-powered productivity coach that helps users optimize their time and focus. It analyzes user activity to provide real-time advice on when to focus, take breaks, and stay on track. Features include app and website blocking, focus music, a flexible Pomodoro timer, and in-depth metrics. </description>
<description> ADHD Numo is a daily planner app designed specifically for individuals with ADHD. It offers features like task gamification, subtask splitting, mood tracking, and daily tips from ADHD coaches. The app aims to help users stay organized, reduce procrastination, and improve overall productivity. </description>
<description> Twos App is a note-taking and list-making app that incorporates AI features. It allows users to create lists, capture quick notes, set reminders, and even generate summaries using AI. The app's focus is on simplicity and ease of use, making it a convenient tool for organizing thoughts and tasks. </description>
<description> Endel is an AI-powered ambient sound generator that creates personalized soundscapes to help users relax, focus, or sleep. It uses machine learning algorithms to analyze users' preferences and create unique audio experiences tailored to their individual needs. </description>
<description> Pi Chat: Pi is your personal AI, designed to be supportive, smart, and available anytime. It can help individuals with ADHD by providing advice, answering questions, and engaging in conversations to help manage daily tasks and challenges. Pi’s conversational and collaborative nature makes it a valuable tool for improving executive function, staying organized, and reducing stress. </description>Now, when we combine this structured knowledge with our structured prompt, the result is nearly perfect. We've effectively guided the AI to produce content that's not only well-structured but also based on reliable, curated information.
This example opens up exciting possibilities for how we can use structured knowledge in education and professional development. Imagine assigning students to research AI tools for people with ADHD, not just listing them, but writing detailed descriptions, user reviews, start-up guides, and tips and tricks
By compiling this information into a structured knowledge base, we're creating a robust, customized AI tool. This approach has several benefits for stduents:
Deep Learning: Students gain in-depth knowledge about their subject matter through thorough research and structured documentation.
Critical Thinking: They learn to evaluate tools and present information in a clear, structured format.
AI Literacy: Students understand how to create and interact with AI knowledge bases, a crucial skill in today's digital landscape.
Practical Application: The resulting knowledge base becomes a valuable resource for the ADHD community, bridging academic work with real-world impact.
Customized AI Interactions: With this structured knowledge base, future AI interactions on this topic will be more accurate, nuanced, and tailored to specific needs.
This exercise demonstrates that by understanding and implementing structure in our writing and knowledge organization, we can significantly enhance our interactions with AI. We're not just using AI tools; we're shaping them to be more effective and aligned with our specific needs and expertise.
As we continue to integrate AI into our writing and learning processes, this structured approach to knowledge will become increasingly crucial. It's a powerful way to ensure that AI serves as a true extension of our expertise and creativity, rather than a generic tool.
Expanding Horizons - Structured Knowledge Across Disciplines
While we've focused primarily on professional writing and AI tools in this post, the potential applications of structured knowledge and AI interactions extend far beyond these boundaries.
Let's consider how this approach could revolutionize learning and engagement in other academic fields:
Literature: Imagine students creating structured knowledge bases for different literary periods, authors, or themes. They could build comprehensive, interconnected databases of character analyses, plot structures, and thematic elements. An AI trained on this structured knowledge could then assist in comparative literature studies or help generate nuanced interpretations of texts.
History: Students could develop structured timelines, cause-and-effect relationships, and interconnected biographies of historical figures. This approach could lead to AI-assisted historical analysis tools that help identify patterns across different eras or regions, offering new perspectives on historical events.
Science: In fields like biology or chemistry, structured knowledge bases could be created for complex systems, reactions, or ecological relationships. AI trained on this data could help students visualize intricate processes or predict outcomes of experiments.
Art History: Structured databases of artistic styles, techniques, and historical contexts could be developed. An AI system could then assist in analyzing new artworks, suggesting potential influences or placing them within broader artistic movements.
Philosophy: Students could create structured arguments for different philosophical positions, including premises, conclusions, and counterarguments. AI could then be used to generate dialogues between different philosophical viewpoints or help students construct robust arguments.
The possibilities are truly endless. By teaching students to structure knowledge in these ways, we're not just preparing them for a world where AI is commonplace; we're equipping them with critical thinking skills that transcend any single technology or field of study.
This approach also fosters a deeper, more interconnected understanding of subject matter. It encourages students to think systematically about how information in their field is organized and related, leading to more nuanced and comprehensive learning outcomes.
As we move forward in this AI-integrated educational landscape, our goal should be to cultivate not just users of AI, but architects of knowledge.
By understanding how to structure information effectively, our students will be better prepared to shape the AI tools of the future, ensuring that these technologies serve as extensions of human creativity and critical thinking rather than replacements for them.
The future of education lies not just in adapting to new technologies, but in actively shaping how those technologies understand and interact with our fields of study. By embracing structured knowledge approaches across disciplines, we're laying the groundwork for a more thoughtful, nuanced, and effective integration of AI in education and beyond.