Have you ever spent way too long trying to figure out how to get soap from the soap dispenser in a public bathroom? 

​​​If so, you understand what poor affordance is.

According to the Interaction Design Foundation, an affordance is "possibility of an action on an object."

We don't hear it said this way too often, but it's like this: An elevator button affords a press. A chair affords a sit. A hyperlink affords a click.

Why am I giving you a vocabulary lesson today?

Because affordance is one of the key interaction design challenges facing the latest wave of groundbreaking technology: ChatGPT.

If you're one of the 100M people who has already tried to use ChatGPT, it's likely that your initial impressions were:

• "Wow, this seems kind of magical!"

• Followed shortly by "What all could I use this for?"

Even the most powerful product can be frustrating and hard to use when it lacks affordance. In the case of ChatGPT, the interface itself is simple to use. But it's less obvious to get real value (the soap!) from it.

Users appreciate guidance. What can I do with this? And how do I do it?

The more a product can do, the more challenging it can be to design adequate affordances. That's part of the reason you've seen so many posts circulating on social media about how to craft a great prompt. These guides are filling an affordance gap.

Over the past weeks and months, we've also seen a wave of niche AI-powered tools and feature announcements. These tools apply the power of AI to specific tasks. For example, products like Midjourney and DALL-E do one thing: they generate images from natural language prompts. By focusing outputs on a more limited sphere, the product instantly becomes more learnable.

My challenge to you? With every product or feature, remember affordance. Make sure that the experience makes it clear what users can do, and how they can do it.

In other words, don't be the frustrating soap dispenser in a public restroom!