How to Use Generative AI, Part 5: Weave

Large Language Models are incredible in their ability to synthesize and recombine vast stores of knowledge, serving as powerful tools for creative and analytical tasks alike. Of all their skills, we think this one is perhaps the most powerful for most users.

By weaving together disparate concepts into novel combinations, these models demonstrate an unparalleled capacity to "innovate", drawing from their extensive training data that spans virtually every domain of human knowledge. This capability allows LLMs to make connections across disciplines that might elude human observers, often bringing to light associations and parallels that spark new ideas or solutions to complex problems. For instance, an LLM might link principles from ecology to urban planning, suggesting a biomimetic approach to city design that enhances sustainability and resilience.

LLMs are useful in reframing unfamiliar in terms that are relatable, using cross-contextual juxtapositions that make complex subjects accessible. You can use them to demystify advanced topics and other complex subjects by drawing analogies to everyday experiences, which broadens your understanding and appreciation of challenging topics. This analogizing power not only aids in education but also stimulates innovation by encouraging you to approach problems from unexplored angles.

By surfacing non-obvious patterns across a diverse array of knowledge, LLMs unveil insights that might remain obscured without such comprehensive cross-referencing capabilities. This aspect is particularly valuable in research and development, where identifying subtle correlations can lead to breakthroughs. For example, an LLM might detect a similarity between the social structures of certain animal species and human organizational behavior, offering new perspectives on leadership and teamwork.

LLMs can play a crucial role in expanding your understanding by translating the complexity of the world into comprehensibility. They do this by recombining information and presenting knowledge in new configurations that highlight underappreciated relationships or simplify intricate concepts. This recombination is not merely a mechanical process but a dynamic form of intellectual creativity. In doing so, LLMs act as intellectual partners that can augment your capacity for creativity and novel idea generation, enabling you to navigate the complexities of modern knowledge landscapes with greater agility and insight.

An LLM is one the best ways to find, what we call, the "outside view". When making a forecast or developing a plan, we can be too narrow in our focus. LLMs act as an expert in a field and can point to good reference classes which help you develop broader (and often more accurate) conceptions of your own problem or opportunity. For instance, if you want to critique your contractor's estimate for a bathroom refurbishment, it's often better to look at variation in other construction projects than it is to deeply critique your specific case.

In essence, the combinatorial and analogizing capabilities of LLMs mark a significant advancement in our ability to process and reinterpret the vast amount of information that defines the contemporary world.

Conversational Scenario: You are a consumer retail company considering launching a new line of athletic apparel. You need to forecast demand to guide inventory planning.

Step 1: Frame the key uncertainties: Prompt the AI to list the most important uncertainties in forecasting demand for the new product line.

Prompt example: I work for a consumer retail company which is considering launching a new line of athletic apparel. What are the key uncertainties we face in predicting demand for our new athletic apparel line to inform inventory planning?

Step 2: Gather reference classes: Prompt the AI to identify 3-5 similar product launches from other retailers that could serve as reference classes. 

Prompt example: Provide examples of 5 consumer retail product launches that are highly analogous to our planned new athletic apparel line launch. Include key data like product type, pricing, market context, and demand.

Step 3: Extract insights: Prompt the AI to analyze the demand patterns from the reference classes and summarize key implications for forecasting demand.  

Prompt example: Review the reference product launches and their demand data. What key insights can we draw regarding likely demand patterns and volumes for our planned new apparel line launch?