Ben Tsai

I’ve been geeking out on espresso recently since learning about its history and how today’s equipment and beans have changed, so the rules-of-thumb I’ve heard about may no longer be valid: what should the ratio be?

In my latest venture towards productivity bliss, I’ve been looking at a lot of different typefaces to fit with my plain text system.

Fraudulent.

I’ve already referenced Addy’s Comprehension Debt piece, but I think it is worth highlighting this quote: Making code cheap to generate doesn’t make understanding cheap to skip.

The paradox: to get better at ai, you need to not use ai.

I recently read this piece on one of my favorite topics entitled On genAI: Was prototyping really a bottleneck?

One pillar of my origin story is when I realized the software I’m building exists as part of a system whose goal is to serve human beings.

Erik Johannes Husom’s piece entitled Outsourcing thinking resonated with me.

I wrote a LinkedIn post: I used to have some regrets.

From Robin Sloan’s blog: Here is one distinction among several: this process can only compound — the models can only “think” by spooling out more text — while human thinking often does the opposite: retreats into silence, because it doesn’t have words yet…

The simulation of judgment in LLMs While models often match expert outputs, our results suggest that they may rely on lexical associations and statistical priors rather than contextual reasoning or normative criteria.

I wasn’t intending to rework my tasks and notes system.

For a long time, I’ve been trying to articulate why interacting with an LLM is so irksome.

Marcus Hutchins (of WannaCry fame) is my new favorite follow.

Should you use an LLM to help you “write?”

A few months ago, I was inspired by this remark from world-class violinist Hilary Hahn, speaking on “natural talent”: At some point, everyone’s got to do the work.

Who is more foolish?

#TheCompoChallenge

Day 1 of using Ugmonk Analog For cyber Monday, I splurged and purchased Ugmonk’s daily and weekly system.

Constraints are liberating.

When we create prototypes, I rarely see a consideration around who the audience is.

My colleague Lindsey M.

Okay, we gotta talk about chatbots.

I’m surprised I haven’t written more about my disdain for AI.

I participated in WeblogPoMo2024 last month.