Probably what went through his head after it went down:

The news just came out about the arrest of the CEO's killer and Polymarket is wayyyyy too quick with releasing their latest betting odds 😂

I use custom meta-prompts for ChatGPT to have it write the way I want it to write. I pasted in the last 15 paragraphs worth of ChatGPT responses from my latest ChatGPT thread into GPTZero and it marked it as 72% human…

Had to try the `calculate_woman_salary` autocomplete that I've been seeing, and it looks like Claude replicates what others are seeing where it autocompletes to less pay for women, but once you include the years of experience into it, the results flip and women are paid more? LLMs are weird man...

I used to get tripped up like this when dealing with the various types of embeds, so I understand how this feels. I think at some point I accepted that I got a close-enough solution, and whenever things like video are added, I'm like "do I REALLY need this in my data pipeline? Hopefully not???"

“Evaluated the ROI on your tech stack” is something that restaurant owners definitely do NOT care about. Contrary to popular belief, there’s a world outside of tech.

it's almost like modern science works so well that people forgot why we invented things like pasteurization or vaccines...

Probably the most well-timed webinar marketing email for me. Coincidentally, I'm (jk, ChatGPT and Copilot) working on an MVP of a two-tower model. Time to smash that register button 😤 (please don't be at 1am PH time...)

"Ugh, I travel too much, I can't decide where to go next"

I mean this is a real problem ig, but one that most people would kill to have. I absolutely love being a digital nomad. Sure, it's not 100% perfect, but they're good problems to have.

Who needs Jira or Trello when you have a ticketing system at home (Notion).

Doing everything by yourself means you end up re-inventing tools that already exist. If I'm at the point though where I need something bigger to track tickets, that's a good problem to have.

I used to hate adding testing and validation and other code quality things because of how long they took but it’s so trivial with LLMs that I really don’t have any more excuse to not add them. Now if my code doesn’t have quality checks it really is my fault 😎

Simple example of how LLMs for coding can give plausible but wrong answers: the error wasn’t syntax-related but rather that the default value for “model_dict” is a global variable, so when you pop the key, the next call won’t have it. Looks like I’ll have a job for a little longer 😎

Not sure what the use of these screening questions for this event was when you could Google these (or even ChatGPT, if you’re even more lazy like me)

tidyverse singlehandedly made many aspects of data science more enjoyable to do in R than in Python, so I'm glad that I can use ggplot in Python instead of googling/ChatGPT-ing "how to bar plot matplotlib" all the time