logo

Welcome!

I'm an interface researcher interested in how technology can support deeper understanding and expand human creative capacity.

I've done research at NYU focused on creativity support tools, developed interfaces at Midjourney and worked as an interaction designer alongside Hugh Dubberly.

With Andy Matuschak, we designed Latticework, a text editing environment aimed to help people synthesize freeform, unstructured documents.

If my work resonates with you, this is a standing invitation to reach out!

matthewwilsonsiu@gmail.com
@MatthewWSiu

Interface Gallery

Bulletin

dec 7 4:45pm
Been thinking about one day moving to the mountains to be closer to nature. I'll invite friends to join me and we'll go hiking, catch up and make good food.

Also just remembered that my logo is a mountain + stream, I had forgotten!

Bulletin

oct 14 12:03am
I've been really enjoying the process of making interfaces recently. It feels like craft to me, like shaping clay or any other material.

Bulletin

oct 5 2:07am
2024 will mark the year that I decided to sit down and master my craft ā€“ digital material design

Bulletin

aug 27 10:45pm
Really enjoy coming back to update this website. It feels reflective to sit down and ask, "do these things still resonate?"

On a separate note, Latticework is up now and am sharing tomorrow ā€“ excited to get it out there.

Bulletin

jun 3 11:09pm
I have a growing collection of interface sketches in my notebook that I want to make.

Stay tuned...

Bulletin

apr 20 1:27am
Assuming we are able to get enough interpretable features across all aspects of the transformer model, how does this change how we can steer the output? What types of ways to edit and explore are possible now?

Bulletin

mar 10 3:17pm
Iā€™m really excited about interpretability. They're developing a new field of neuroscience for artificial minds.

Compared to biology, the data we need is all there in the activations and weights. If only we can develop the methods and tools to see it.

Bulletin

feb 20 9:47pm
Mechanistic interpretability is fascinating and has a surprising amount of overlap with my interests

By understanding these models better, we may be able to learn new ways to think about the world, and ourselves
Caret
Caret

Places

Index