Is It Me or the Tool? Exploring AI’s Hidden Menu Together
“Is this tool not working for me, or is how I’m using it not working for me?” has always been a difficult question for me to answer with tools that have steep learning curves. How much time do I spend learning before deciding to move on?
With AI tools, there’s the added ambiguity of not knowing whether my experimentation is leading toward a sustainable workflow. Even when I figure out the next step, I don’t know whether it’s bringing me closer to my bigger intention — because there’s no map of what’s possible or how the pieces fit together.
Get Enigmas Next Door (aka Tara Raj)’s stories in your inbox
Join Medium for free to get updates from this writer.
I started my career as a software engineer, where I saw a lot of bias toward treating friction as user error, even when the “right way” to do something was never clearly documented. For a few years, I got paid to navigate this friction on the job, and I felt for customers who had to navigate it on top of their own jobs. That was part of what eventually led me to transition into UX research.
This ambiguity can also obscure systemic barriers. When someone else has a better experience with a product, how do we know whether it’s because they’ve put in more effort, found a different approach, or because the product carries more of the load for them — because it’s optimized for people who experience or process the world more like they do?
This is especially difficult to see when a product produces an output for everyone. When a webpage doesn’t load with a screen reader or a translation tool says “translation not available,” there’s at least some indication of what’s inaccessible. AI chatbots don’t necessarily signal when we’re missing out. They might tell us something isn’t possible or fail silently, and we may not discover that better is possible until we hear about someone else getting support from AI where it failed us.
When tech doesn’t publish the menu of what’s possible, discovering it requires learning from each other. And when we’re using the same interface across different contexts, there’s an opportunity to learn not just what works for others, but what about their contexts might explain why it works — or doesn’t. That gives us more to go on when figuring out what might work for us.
