DeepSeek and design

“Of course, users can learn over time what prompts work well and which don't,” wrote Amelia Wattenberger, a research engineer at Github, in a 2023 post about the limitations of chatbot design. “But the burden to learn what works still lies with every single user. When it could instead be baked into the interface.”

What an astute observation from Casey. I was talking with a friend recently about this exact problem. He called out how the input field is such a limiting design pattern, and ultimately holds back so many people from engaging with, and benefiting from AI chat bots.

DeepSeek responds to your queries by telling you what it thinks you want, and what it thinks it should do in response.

I think there’s still an element we haven’t arrived at yet, where the user interface could be generated on the fly, as needed by the AI, to help the customer solve their problems in the moment. WIth that said, having an AI bot that tells me what it’s understanding, then offers guidance, is a huge step.

If I can understand the thought process behind a person’s decision, then even if I disagree with the conclusion, I still benefit from steps taken to reach that conclusion. I’m excited to see personally if this reasoning helps me engage better with chat bots.

Via Platformer.