Generative AI is becoming a powerful tool to help address a wide range of business problems in creative ways. However, areas for applying generative AI are broad, and many uses diverge from how we work today.
This struggle to understand AI’s potential is causing most of us (myself included) to get stuck in beginner mode, unable to see how the technology might be applied inside our organizations. What I needed was a better understanding of what generative AI is good at. I hoped this understanding might help me imagine how to put the technology to work in relevant and meaningful ways. (I also wanted to help my fellow participants at InnoLead’s upcoming “Leveraging A.I. in Business” workshop think about how we could use the technology as part of our prototyping work there.)
I began by creating a list of use cases where generative AI can currently solve meaningful business problems. These examples were helpful and provocative, but involved specific industries and unique business problems so were difficult to generalize.
What I needed was a way to look across use cases, to see a set of generalizable patterns that can be applied across industries and roles. By looking at these patterns, I might more quickly reveal where generative AI might apply to my work in meaningful and relevant ways.
The next step? Enlist generative AI itself in evaluating my use cases to see what it came up with for generalizable patterns. Using ChatGPT, I created a prompt and supplied my list of use cases. The following is ChatGPT’s response. (At the very bottom of this piece is the original prompt and list of use cases I used for context.)
I’ll let you evaluate the ability of ChatGPT to see patterns and the usefulness of these patterns in helping inspire where generative AI might apply to your own organization.
ChatGPT’s Response:
“Automating Repetitive Tasks: Generative AI is particularly effective at automating repetitive tasks such as generating names, words, ideas, themes, and story fragments. It can also assist in generating code or other partially repetitive or predictable content such as business proposals or contracts. The AI can suggest text as you begin to write or provide structured inputs and guidance. This pattern is also evident in editing and packaging content, such as video footage, photos, or written texts, and combining or editing them into finished products like video reels, slideshows, or presentations that explain or market products and services.
Personalization: Generative AI can be used to provide a high degree of personalization. It can modify a product or service to serve the unique needs or preferences of a customer. For example, it can instantly update on-screen text to align with a customer’s preferred style or update a product’s features or colors to align with what the customer needs right now. It can also serve as a personal shopping assistant or recommendation engine, suggesting items based on what it learns about the customer.
Augmenting Human Decision-Making: Generative AI can augment human decision-making by serving as a personalized marketing solution. It can generate highly targeted marketing output at scale for niche audiences, allowing marketers to create significant amounts of personalized content for low cost to target smaller segments of the market. Different marketing outputs can even be tested using A/B tests and optimized dynamically from the results. Additionally, generative AI can be used as a customer support chatbot, understanding customer queries, providing relevant responses, detecting patterns, and escalating when necessary.
Diagnosis and Discovery: Generative AI can be used as a diagnostic assistant for mechanics, technicians, or healthcare providers. It can take symptomatic inputs and assist with initial diagnoses or suggest further data or tests. It can also be used to review transcripts or other large data sets, looking for identified patterns or counteracting human bias by looking for counterfactuals.
Sensemaking and Analysis: Generative AI can be used for sensemaking, such as doing content analysis, sentiment analysis, or translating languages. It can explain things the user doesn’t understand and pull insights from transcripts, customer reviews, or other large data sets. Additionally, it can assist in the design process by rapidly creating design assets for user feedback, generating multiple options, and optimizing solutions based on feedback. This allows designers to speed up design cycles and get better feedback sooner.
In summary, generative AI is most effective at automating repetitive tasks, providing a high degree of personalization, augmenting human decision-making, aiding in diagnosis and discovery, and facilitating sensemaking and analysis. These use cases demonstrate the broad range of applications for generative AI, highlighting its ability to enhance and support human creativity, decision-making, and productivity across multiple industries and domains.”
| My Initial Prompt and Use Cases |
| Following this prompt, I will submit a list of use cases for generative AI. I’d like you to analyze this list of use cases as if you’re a consultant from a research-oriented consulting firm. Provide me with generalizable patterns for what generative AI is good at that are not specific to an industry and can apply to different jobs. Think of these patterns as categories for how to apply generative AI, for example, that AI is good at handling large numbers of submissions. Label each pattern with a descriptive, multiword label and describe what use cases might fit under this label. Complete your analysis with a short paragraph summarizing what generative AI is most effective at addressing without just repeating what was said previously. When you understand and are ready for my list of use cases, say “I understand and am ready.” Latest list of use cases: • As an input to the creative process, generating names, words, ideas, themes, story fragments and other creative inputs to build upon. • As a suggestion tool for those writing code or other partially repetitive or predictable content, like business proposals or contracts, where the generative AI can suggest text as you begin to write or when you provide structured inputs and guidance. • As a personalized marketing solution, able to generate highly targeted marketing output at scale for niche audiences. For example, significant amounts of personalized content can be generated for low cost in order to target smaller segments of the market, e.g. the over 55 urban Pekinese owner who might buy your dog food. Different marketing outputs can even be tested using A/B tests and optimized dynamically from the results. • Edit and/or package content, like video footage, photos or written texts and combine or edit them into finished products like video reels, slideshows or presentations that explain or market products and services or tell stories, put existing content into new formats, etc. allowing writers and producers to produce more output or to reduce the time they spend producing content. • To review transcripts or other large data sets, looking for identified patterns or for counteracting human bias by looking for counterfactuals. • Customer support chatbots that can augment humans by understanding customer queries, providing relevant responses, detecting patterns, and escalating. • Managing all sorts of human submissions, such as consumer complaints, bug reports or applications for grants or jobs at scale. • Generative AI could also be used as a bias checker to keep humans from making errors when evaluating these submissions. • Diagnostic assistant for mechanics, technicians or health care providers, taking symptomatic inputs and assisting with initial diagnoses or suggestions for further data or tests. • Writing or checking software code. • Assisting in the design process by rapidly creating design assets for user feedback, generating multiple options and optimizing solutions based on feedback. For example, generative AI could serve as a tool to work early on with a designer who presents several different directions for design elements, colors and so on based on basic inputs. This then gives creative ideas, interesting directions or rules out options the designer can then build on, permitting the designer to speed up design cycles and get better feedback sooner. • As a personalization solution, where knowledge about the customer can permit a generative AI to instantly modify a product or service to serve the unique needs or preferences of a customer, like updating on-screen text to align with a customer’s preferred style or to update a product’s features or colors to align with what this customer needs right now. • Providing advice, either as an input for an advisor, or as an alternative to an advisor. • Acting as a personal assistant, such as a shopping assistant or recommendation engine suggesting things that I might need or selecting from many choices based on what the generative AI learns about me. The generative AI might even make low-risk decisions on my behalf, such as scheduling meetings or ordering products. • Finding answers that are more personalized to my needs than a simple search engine. • Sense making, such as translating languages, explaining things to me I don’t understand, pulling insights from transcripts, customer reviews or other large data sets, or doing content analysis, such as sentiment analysis. |
Keith Hopper is Founder of Danger Fort Labs, Keith has helped design courses and taught Product Discovery at Harvard Business School and at Olin College of Engineering. He has also worked closely with dozens of startups and was voted Boston’s Most Valuable Techstars Mentor. Previously, he was at NPR, where he ran Product for the Digital Services Division.
