Trainer, Consultant & Nonprofit Innovator in digital transformation & workplace wellbeing, recognized by Fast Company & NTEN Lifetime Achievement Award.
522,246 followers
In January, we found ourselves with a pleasant surprise – some flight credits were expiring, and we needed to decide where to go. We brainstormed with our human brains, but our list consisted of our usual suspects. That's when I decided to try something new. In January, 2024, OpenAI launched Custom GPTs, including one built by Ethan Mollick called "Innovator GPT," a tool designed for structured brainstorming. Before creating my own GPT (and I've created a bunch of them), I wanted to test one designed and built by an expert. I shared our travel problem with the GPT, "Where to go?" I also included context details such as timing, trip duration, weather preferences, preferred activities, and flight duration. The GPT then performed a series of structured ideation activities and presented me with a list of 40 destinations. Sydney was the only Australian destination on the list, but it sparked ideas for Tasmania and Melbourne. Next, I brainstormed with the GPT about what to do in Melbourne and Tasmania: 1. Open-Ended Question: Designing a week-long itinerary combining Melbourne and Tasmania's attractions. 2. Idea Generation: Provided 10 ideas, then added another 10 and compiled them in a Word document. 3. Constraint-Based Ideas: Added constraints to refine ideas, like "What would an astronaut do in Melbourne?" 4. Idea Combination: Paired ideas to generate new ones. 5. Final Selection: Ranked the top five ideas based on feasibility, impact, and originality, presented in a chart and Word document. While this process offered fresh trip ideas, it didn't create a specific itinerary. For that, I used Microsoft's Co-Pilot (creative mode) to gather real-time information on hotels, flights, restaurants, hikes, stationery stores, and more. If only I had an intelligent agent to handle all the bookings, but I know that is coming with ChatGPT 5 and the next wave. We also consulted our favorite travel sites and, truth be told, checked out travel guides from the local library. And talked to other humans about their experiences in these destinations. This was for fine-tuning and fact-checking. Now, we're a little over halfway through our 15-hour flight from the Bay Area to our unexpected destination: Melbourne & Tasmania. We can't wait to see what adventures await us as part of the trip co-designed, and co-planned with generative AI. In terms of organizational adoption of generative AI, some points: * Test-drive a well-designed custom GPT or Pilot created by an expert to understand what it can do. * Think through your workflow for the task and explore the points of integration between human skills and LLM skills. * Keep humans in the loop. I think intertwining is the way to go with humans always having the first and last review.