Chris Hemphill

['data science', 'health tech', 'running', 'synthesizers']

not funny enough for twitter, but too edgy for linkedin. opinions are my own. (they/them)

[Chris Hemphill @luketrailrunner]

Are #datascience jobs safe?

https://t.co/oqToWeydxp

[Chris Hemphill @luketrailrunner]

@charliermarsh @tdhopper uv and ruff now required for package sharing at my biggest client… and they’re a health insurance company!

[Talia Ringer 🕊🪬 @TaliaRinger]

It was a mistake to call Chain of Thought by that name, because now people think it actually represents a thought process rather than some thought-inspired intermediate output that affects final output in ways we don't yet fully understand, sometimes better when it is nonsense

[Chris Hemphill @luketrailrunner]

@Chris_Orlob What are the external costs of not trynna get the bag?

[Chris Hemphill @luketrailrunner]

@jodiecongirl That plus their slop glorifying algorithm has me wondering who took their data science south.

LI, for like 3 or 4 good years, used to be a delightful place

[Jodi Beggs @jodiecongirl]

i am convinced that one of their analysts told them that more active users had higher LTV is some way and the company blindly decided this relationship was causal and tried to pull the activity lever

[Chris Hemphill @luketrailrunner]

@AllenDowney How did it go? Should we be expecting “Think Causal” sometime soon?

[Simone Conradi @S_Conradi]

Another one!
Made with #python #numpy #matplotlib and #marimo (and also #wigglystuff)

[Chris Hemphill @luketrailrunner]

@timnitGebru Amodei and Altman tend to sound like this to me -

p(doom) = 0.25
p(doom | do(make_me_a_trillionaire)) = 0

[@timnitGebru (@dair-community.social/bsky.social) @timnitGebru]

Remember folks, they're all about "ethics" and "safety" this money is just the cost of bringing us utopia through beneficial "artificial general intelligence" and preventing the bad guys from creating the machine devil that will kill us all.

https://t.co/VlZZ7yGg2h

[philip lewis @Phil_Lewis_]

Louisiana passes new congressional map, dismantling one majority-Black district https://t.co/PmlsrsYY5C

[staysaasy @staysaasy]

AI is weird and mathy, so people love using their knowledge of AI to flex on others. Peer pressure is a bitch so we've hit the point where questioning any aspect of AI's usefulness or ROI makes you look like a depraved luddite who thinks that the sun revolves around the earth.

[エッグサラダ @synthicyde]

Shroudies don't even respond when you ask them what genre death's dynamic shroud is anymore, they just look at you like this

[Science girl @sciencegirl]

Transformations of life

📹 ilyas_insects

[Chris Hemphill @luketrailrunner]

@dieworkwear Nonbinary is so much more fun tbh

[derek guy @dieworkwear]

weird how masculinity is innate and natural but there are a million accounts on here dictating the specific ways masculinity is allowed to be expressed

[Distributed AI Research Institute is on Mastodon @DAIRInstitute]

🧵

[Chris Hemphill @luketrailrunner]

Made this for some friends in a book club (Causal AI by Robert Ness), but figured more folks oughta see how @marimo_io forces good behaviors in notebooks.

This is a teensy tiny bit of why I like it. Worth looking at if you prototype in notebooks and want tight, transferable code

[Nikhil Krishnan @nikillinit]

Since Out-Of-Pocket started in Feb 2020, I've written 267 posts! Approximately once a week for 6.5 years, and ~1800 pages total

More than anything, I'm proud of the posts. I've never really used clickbait or current events, I've always tried to give nuanced views of healthcare issues, and I try to put out ideas on how we can fix the healthcare system.

And also...I have basically used new meme formats for every post. You have no idea how difficult that is lol, honestly probably the hardest part.

I've been told I need to be more celebratory in the milestones. Keeping a writing cadence is hard, but I've enjoyed every minute of it.

[Chris Hemphill @luketrailrunner]

@deliprao @natolambert It means great, awesome ethics, right?

[David Lieb @dflieb]

Founders overestimate how hard it is to become a top 100 expert on most subjects.

For most things you need to solve, there’s usually a very clear path: just go do the work, obsessively, and become friends with everyone else doing it.