Can we outsource empathy?
Three years ago I ran an exploratory workshop. My goal was to learn how AI could impact different parts of the creative problem solving process.
That workshop, which I referred to as HMWAI ("How Might We AI?") involved two separate groups, one where AI was used and one where it wasn't. Other than that, the workshops were the same. AI has advanced tremendously since then, but much of what I observed then I still observe today.
When the HMWAI workshops were over, I interviewed each of the participants about their experience and how they felt at each step. Below is how the teams approached each step and the feedback they shared after the workshop about that step.
Step 1. Empathy Building
Approach:
The AI team chose to do research through prompts. They found lots of information about similar challenges and what's been done about them. They worked quickly and had AI summarize what they found into a cohesive point of view about the problems and where the opportunities were.The non AI team spoke with customers. They did so in pairs, with one person leading the conversation and the other capturing insights. They sorted the insights from all conversations into themes which helped them form a point of view about the problems and opportunities.
Feedback:
The AI group expressed that although they wanted to solve the problem, they didn't feel very connected to it. They felt confident AI was up to the task.The non- AI group expressed that they really cared. The empathy building step had helped them feel what it was like for the customer and motivated them to make things better.
Each team took their point of view into the next step.
Step 2. Creative Ideation
Approach:
*For both groups I asked them to first work separately from their teammates to generate some ideas then come together to narrow down on one idea to take forward.
The AI team worked at lightning speed. They each fed their interpretation of the problem as a prompt and asked for examples and new ideas for how to address it. They used additional prompts to refine and prioritize what they received. When they came together, they fed all of the ideas back into AI to help prioritize which one they would align on for prototyping.The non-AI group drew upon the insights from the empathy building step, used google and their own lived experiences to find analogous examples. They worked separately at first to gather lots of ideas, then together to align around an idea they felt was most connected to their point of view about the problem to be solved.
Feedback from participants:
The AI group described this process as fun and easy. They each came up with similar ideas so it was easy to narrow down to one. There wasn’t a whole lot of diversity, but they trusted AI to generate good ideas.The non AI group said they felt grounded in the point of view they'd aligned around. It helped them find analogous examples. They felt the process of working separately first helped them come up with a diverse set of ideas to narrow down from.
Step 3. Idea Prototyping and Testing
Approach:
Much like the prior step, the AI team worked fast. They used different AI tools to generate mock ups and process flows of the idea. They used AI to create a script for testing the idea. After the tests, they fed all the testing insights into AI for summarization.The non-AI team sketched ideas on paper, then used an AI-less design software to refine the idea. They worked as a team to come up with a script for testing. During testing, they each captured insights as they observed, then came together as a group to pick out what was most interesting.
Feedback from participants:
The AI group appreciated how fast they could create and iterate on ideas. They felt confident they'd be able to do multiple rounds of testing and make changes after each test based on the learning prompts they gave it. Users thought the ideas were cool, but didn’t really see how they related to their own experiences and needs.The non-AI group felt like they were able to do a good job of representing the idea in their prototype, but worried about making quick iterations between tests. Users saw a clear connection between the solution and their problem.
Since that workshop, I've continued to experiment with how and where to leverage AI. Given what I've observed, this is the model I use today.
Empathy Building: This one has to be all human. Using AI for this doesn't just limit empathy. I've watched teams fully detach and just not care. It becomes AI's problem to solve, not theirs.
Creative Ideation: AI is a great compliment to human ideation and creativity, but it must be a complement, not the only method.
Idea Prototyping and Testing: AI prototyping tools are incredibly powerful. You can input a prompt of just a few sentences and minutes later have a prototype that represents a process, an interface or even an expression of a service. Humans should still conduct the testing to refine their empathy for the customer, but new learning can easily be fed back into the AI prototyping tool.
Artificial Intelligence adds a ton of value to many parts of the creative problem solving process. The biggest pitfall for companies is around empathy building. If teams inside organizations are outsourcing their empathy, it leads to them outsourcing their creativity. That limits innovation. If people don't care, there's no incentive to get creative. This is a huge business problem. Do not outsource empathy.