How should I use AI?
An old colleague and friend of mine was an excellent photographer. One day, he showed me his website. He had beautiful photos of family, friends, nature. Every one of them looked professional.
“What’s your secret?” I asked.
“It’s all about the camera and the lens.” He said.
This was 2010 and the iPhone camera was still pretty bare bones - incapable of doing anything close to what he was showing me. I picked his brain some more. What kind of camera? What kind of lens?
He had a wealth of photography knowledge and recommended a sleek looking Nikon D5000 with a Tamron lens. I clicked the buy button and waited eagerly, thinking about all the amazing photos I would take.
When the camera arrived, I carefully opened the box and held the camera in my hand. It was heavy. The kind of heavy that means something is really good quality. I attached the lens and the shoulder strap. I put it around my neck and pretended to snap away like a fashion photographer.
I examined it more closely. There were dials, buttons and knobs. All of them were labeled in a language I didn’t understand. I couldn’t wait to learn how to use it. Excited about my new toy, I put everything into the travel case and placed it on a shelf in my closet. Tomorrow I’d start watching Youtube videos and taking practice shots.
A week went by. Then another. Every time I promised myself I’d start to learn, I lost motivation. Sometimes I’d take it out of the case and take some pictures. I adjusted the buttons and knobs, clueless about what they did. Along with not knowing how to operate the camera, I also couldn’t come up with a reason why I would operate the camera. Other than, “take better pictures”, I had no real use case to get me started. 15 years later, my Nikon D5000 is still sitting on my shelf, collecting dust. What happened?
What if I’d bought the camera for a more specific purpose? What if my main use case was something like taking photos of my kids during swim class at the poorly lit local pool? Surely there was a setting somewhere for low lighting that could capture action shots without being blurry. A use case like that would have given me specific motivation to start. I could have learned how to solve just that problem. I bet that would have led to me learning more.
That feeling of not knowing where to begin or why feels very similar to what I hear from many people about AI. People have been told by their companies to use AI to make things more efficient. “Find all the tasks you can do with AI!” Leaders have invested millions in new tools but remain quietly frustrated about low adoption rates amongst employees. People stare at the ChatGPT prompt box, unsure what to type.
Most companies attack this problem through training and mandates. Training is definitely important, but is there a missing piece? What if we found specific reasons and use cases first, then moved into targeted training to address just those use cases?
It seems overly simple to order things this way, yet the approach is often rejected. “Too time consuming” or “We’ll end up using AI anyway” are some common reasons.
What if there were a simple way to find the use cases? What if you could do it in 20 minutes?
Here’s a method I’ve tried. It won’t help you merge millions of data points or develop powerful predictive models, but it will help people get unstuck.
Step 1: Unpack
Write down all the tasks you do - individually or with a team. Use stickies or a virtual tool like Miro or Mural. One task per sticky. For this part, you can look at the calendar over the last week, project databases, to do lists - anything that helps you remember all the things.
Step 2: Plot
Create a quick 2X2 grid with effort on one axis and return on the other - each one going from hight to low. Plot each task where you think it belongs on the 2X2. Work quickly and don’t worry about being perfect.
Step 3: Categorize
Once all the tasks have been plotted, overlay these categories on each quadrant. You should have the following quadrants:
high effort, low return
high effort, high return
low effort, low return
low effort, high return
Step 4: Review the grid
You should now have some tasks that can be eliminated, some that can be left alone and some that are prime candidates for AI. Instead of just finding tasks, you will have taken things one step further and found use cases.
An exercise like this might have helped me learn how to use my camera. Who knows? Maybe it would have shown me that I didn't really need a camera, but something else. Wherever it led me, it was probably worth 20 minutes of my time.
What do you think?