For all of the mind-boggling new capability, power, and potential that the latest generative AI models have unlocked, we’re also met with a new/old comical irony: we once again have the “blank page” problem.
As ChatGPT, Claude, MidJourney, and other generative AI models and apps have burst onto the scene, they’ve shown us how software algorithms can now help us kickstart processes and eliminate the painful but very human part of starting work – the bit where we stare at the “blank page” and contemplate our existence, our role in the universe, etc. The part where we bang our heads against the desk to come up with the first few characters, colors, shapes, or concepts for what needs to be produced.
However, the example use cases that are meant to get our juices flowing are instead the softballs that don’t quite cut the mustard for knowledge work in the enterprise – e.g.: “let ChatGPT plan your trip across Spain” or “help me write a poem about X”.
But I need Gen AI to help me accelerate the execution of real work, not just vacation planning!
So alas, we’re met with the proverbial “blank page” of coming up with use cases for generative AI.
While generative AI often seems magical and sometimes even human, it is not. These are machines. Working with machines requires us to apply structure to our thoughts and processes to effectively integrate them. While this is hard, consider it a leg up that we humans still have on the machines, and cherish it!
Simultaneously, we should all lower our expectations for generative AI – as of now, these models are NOT human-like automatons with the world’s knowledge and capability baked into their “brains”. They are still machines that require good instructions and need us to work with them in a structured fashion.
So, let’s apply some structure to Generative AI use case identification.
Before jumping into the key questions, consider this: Generative AI is superior to humans in two key ways:
It can scale horizontally (e.g., can do many tasks at once)
It is fast (e.g., can write a 500-word email newsletter draft in seconds)
There are other advantages as well, but to focus us on use case application, let’s keep it to the above for now.
Given these advantages, and remembering that we’re dealing with a powerful yet still disadvantaged MACHINE (not a human in a computer box), we should ask ourselves these 3 key questions to surface potential generative AI use cases:
And for the more visual folks, the 1-pager looks like this:
As we step through the process of asking these three questions, we’ll arrive at a place where we can start to see where generative AI can add value to our work.
While ideas percolate as a result of this process, it’s helpful to then take these use case ideas and visually plot them on a 2x2 matrix.
As my team and I build out our own generative AI use cases and consult with clients on the same, we’ve found a helpful structure to apply is ideating Generative AI use cases across 2 axis:
Employee Facing to External Facing (x-axis)
Content Creation to Decision Support (y-axis)
An example matrix populated with sample use cases looks like this:
Visualizing where the use cases fall within this framework often helps spur on added ideation and highlight the potential applications and risk of the use case.
While we won’t delve into managing generative AI risk in this post, it is a crucial part of the generative AI use case lifecycle.
While the hype for generative AI is massive right now, figuring out where and how to apply it is often a challenge – especially since so much of the hype may make the technology out to be much more of a magical autonomous analyst than it actually is.
But if we ground our thinking in the structure necessary to effectively integrate it, the potential upside is massive. Generative AI can execute some tasks at multitudes of scale and speed that humans can, which frees us up to do more valuable and interesting work.
Disclaimer:
This blog post was written entirely by me (Phil Marsalona), a HUMAN – I promise!