

Discover more from Cyborgs Writing
Bridging AI Operations and Human Expertise with Tailored Frameworks
Recipes for getting AI to think
Taking a structured approach to prompting is a lot like developing gourmet dishes.
Yes, you need to experiment in unstructured ways, but as you see how flavors, textures, and aromas fit together, you have to record, organize, and track your process.
You know … a recipe.
Generative AI models don’t think … we have to help them think.
In prompt operations, that’s called a framework.
Frameworks aren't just about guiding the AI; they're about refining the AI's focus to get more tailored and precise outputs.
Generative AI models don’t think … we have to help them think.
This is an adapted excerpt from my courses on prompt operations, currently in development.
Why Frameworks matter.
Frameworks lie at the heart of human problem-solving and content creation. These structured methodologies guide our thinking, help us organize information, and break down complex problems. They act as cognitive roadmaps, providing clarity and direction in many situations, from business decisions to educational strategies.
As we navigate new GenAI technologies, these frameworks are key in directing AI. Even though AI tools like ChatGPT are designed to project reasoning, LLM technology doesn’t really reason like humans without additional input. They mostly function off of linguistic probabilities and fine-tuned training. While this can get us good results at times, it is important not to mistake this for thinking.
If you want AI to use a specific method, you have to guide it yourself.
Just as frameworks have shaped our cognitive processes, they can also shape AI's "thought" patterns, ensuring more aligned, relevant, and insightful outputs.
Asking AI to analyze or reason definitely has its uses, but when you want to manage your AI output in more specific ways, frameworks need to be embedded in your prompt management.
What Frameworks?
Industry-specific frameworks are key to developing expertise and solving problems that have no easy or formulaic solution. They help specific sectors focus on what truly matters to them and translate expertise into actionable strategies.
Too often, these ways of thinking are implicit. But just like when you are training an intern or new employee, you have to make these explicit if you want AI to produce the output you are looking for.
Just look through all the “alphabet soup” in your own field or expertise … we just love our acronyms!
Here are just a few examples across various sectors:
Innovation: Methods like design thinking aid the creative process. They guide brainstorming and foster innovative thinking to design user-centric solutions.
Business: From SWOT and PESTLE analyses, frameworks help businesses evaluate scenarios and draft strategies with precision.
Project Management: Tools such as the Agile, Gantt Charts, and the Waterfall model offer a structured approach to plan, execute, and control projects effectively.
Teaching: Tools like Bloom's Taxonomy and Backward Design help educators craft effective learning objectives and strategies.
Content Creation: Narratives like The Hero's Journey, marketing structures like AIDA, and the SCQA method are essential for crafting compelling content.
Tech Writing: Organized structures like API Documentation, Use Case Scenarios, and SOP Templates ensure clarity and consistency in technical content.
Healthcare: Diagnostic frameworks like SOAP (Subjective, Objective, Assessment, Plan) and SBAR (Situation, Background, Assessment, Recommendation) contribute to more effective communication between healthcare providers and accurate patient care.
Communication: Persuasion techniques like Ethos, Pathos, and Logos, or the Rhetorical Triangle, provide a basis for formulating persuasive arguments and compelling messages.
It is true … AI can already deploy many of these frameworks from its data set … but it can’t necessarily deploy these frameworks in ways specific to your organization or team. You and your team need to work together to further define these frameworks so that you can structure your prompt operations in more specific ways.
While these frameworks serve as a starting point, we bring to the table our own understanding of complex situations, empathy for different perspectives, and the ability to adapt these structures to the unique contexts of our problems.
Your team should refine these frameworks based on the nuances of the situation, your knowledge of culture, and your intuition about the most effective approaches. This dynamic human presence will always be crucial for the successful deployment of AI.
By taking these broad frameworks and refining them, tailoring them to the unique needs and intricacies of your organization, you can guide AI to produce outputs that resonate more deeply with your specific goals and contexts.
Example 1: Social Media Marketing
Social media marketing is a great place to think about how to organize and structure prompts using frameworks.
Imagine a marketing team at a children's book publisher wanting to create social media content to promote new book releases. They decide to use the AIDA framework (Attention, Interest, Desire, Action) to craft their posts and ads.
However, they realize that the classic AIDA model needs to be adapted to resonate with their specific audience - busy parents of young kids. So they refine the framework in this way:
[Attention] Use bright, punchy language with a clear metaphor that is relatable to both kids and parents.
[Interest] Highlight characters and themes relatable to family life that pique curiosity. Pose a teaser question to keep parents reading.
[Desire] Communicate the benefits of reading together - bonding time, literacy development, restful routine.
[Action] Reduce friction by highlighting the link to a sales page. Suggest simple activities related to the story for parents to do with their kids.
By tailoring this content marketing framework to the needs of harried parent, this prompt is more likely to craft social posts that resonate more deeply. If not, the team can adapt the prompt and easily swap out blocks for other audiences and contexts.
Example 2: Design Thinking Framework
One of my favorite examples of how frameworks function is in design thinking. While design thinking helps us structure our problem-solving process, there is no one way to deploy this method. It can be different every time.
Organizations also may have different approaches to design thinking that are not inherently obvious in the framework itself.
Let’s take my coaching with District C, a non-profit organization that takes design thinking to the classroom through what they call a teamship. District C coaches frame various design thinking mindsets and tools in ways that makes sense to students. The goal is to help students adopt these frameworks in group problem-solving without constantly telling students what they should be doing.
Something we call “poking the raft.”
Over the summer, I’ve found AI a useful collaborator … both as a coach and an aid to my students. But you can’t just ask AI to solve a problem using design thinking. It will simulate the process … but not actually go through the process. And it won’t necessarily deploy design thinking in a way that matches District C’s approach or the specific contexts around the design challenge.
So before you use AI in a District C project, you have to give it the right framework that includes the specific context. Who are the students? Business partners? Problems? How are you defining the elements of the design thinking process?
This is hard work … but once you have these prompt blocks worked out, they can be reused, adapted, and revised as these contexts and frameworks change.
Here is just one small example.
One of the most important parts of the District C process is coming up with an insight and tying the prototype to that insight. Students arguably spend the most time leading up to this insight. Instead of just asking AI to give feedback on an insight, I first define important aspects of the design thinking framework.
[DEFINITION] An insight is a new way of thinking that helps us redefine the root problem so that we can build a solution for our partner. It should include an emotional or motivational element.
[FORMAT] Use this sentence structure: -Client- thinks the problem is [blank], but really the problem is [blank].
[TASK] Give me 10 different versions of this insight.
###
[INSIGHT] -Write out current version of insight-
Students often have great insights, but have difficulty articulating them. Instead of asking AI to give us an insight, we are asking AI to rework our specific insight within District C’s framework.
One might think that students will then just pick one that sounds nice … cut and paste. But by this time, students have learned important methods for collaborating, and they treat AI output as a collaborator.
For example, they might spend 5 minutes coding the insights … a District C tool that trains students to annotate notes as a team. Or they might use the “like tool” where they vote for their favorite version (along with their own versions). Or they might take a solo flight to rewrite new insights on their own based on the output.
There is no one way to collaborate with AI! Once students realize this, they are less likely to have AI simply take over for them.
The AI's job isn't to replace human expertise, but to augment it. This requires clear thinking, clear communication, and collaboration.
All skilled chefs have recipes … but they also know how to adapt and experiment with those recipes in new contexts. That’s the expertise you and your team brings to prompt operations.
Whether it’s in business, teaching, or any other sector, frameworks bridge the best of both worlds, helping AI operations become not only efficient and accurate but also human-centric, empathetic, and adaptive.