This lovely paper from isolates a nice pattern for designing ML-enhanced interfaces (IA, not AI): design a shared representation for domain actions, so that suggestions can be fluidly/incrementally adapted alongside user-initiated action. pnas.org/content/pnas/1
Conversation
Replying to
The paper includes several examples, but the translation interface is my favorite—some of the most thoughtful academic HCI design work I've seen in the past few years. Nice work, Spence Green et al (now commercialized as ) idl.cs.washington.edu/papers/ptm/
12
translation is a great example, I remember first experiencing type-ahead in google search and then loving how it evolved into email tab-completable suggestions & @ tagging people.
Looking forward especially to how this paradigm evolves beyond text / natural language.
3
4
It's been interesting to see GTP-style language models assist in the generation of figma designs, code snippets, music, images, video, and others outputs derivative from natural language.
1
Replying to
Thank you! (It helps to have wonderful collaborators…) I also gave a talk about this work back at NeurIPS 2019. Here is the video:
1
8
Show more replies



