SWE+TPM working on AI and nuclear stuff. Lover of music, public health, and spicy food. Opinions my own.

D.C./Baltimore Metro
Joined March 2007
Michael Pipolo retweeted
Anglicized Native American names are by far the best place names in the US. If we're going to go around randomly renaming things we should be switching to those. Consider how much better Manhattan is as a name than New York, or how much better Kentucky is than West Virginia.
20
42
9
2,030
37,988
I think AI will be a transformative technology like the internet by which I mean it will buy us another 2-3 decades of 2.5% GDP growth.
41
81
19
2,178
95,681
Let us imagine that RSI is achieved in mid-2027 and we have an AI capable of melting any networked information system on the planet in 2028. What are the chances that this administration doesn’t do everything in its power to point that at 2028 election infrastructure?
59
Feeling incredibly sanguine about our chances when some of the folks most worried about AI wiping us all out also believe that U.S. GDP growth will reach 17.1% in 2030. anthropic.com/research/ai-ex…
1
59
You think you have a procrastination problem? I’m still avoiding replying to this one text from July 2025. Probably time to let go and mark it as read.
1
45
A bit perplexed by You Can See Everything discourse. Why is this specific documentary getting such outsized attention? Didn’t realize Elizabeth Holmes was of such interest to America. You’d think it was a blockbuster from my local theater’s showtimes.
1
90
Michael Pipolo retweeted
You either live as a supporter of the umlaut or you diacritic
28
251
12
2,366
38,702
Watching AI progress, I'm starting to wonder if my somewhat niche expertise in platform work for software teams building across airgapped systems might turn out to be an edge.
1
38
I can’t decide whether the writing in Blood Meridian is good or completely ridiculous. Like, is he for real with this.
1
52
30,000 merges a day and they still emailing us magic links to log in to claude dot com. Dear lord.
New on our Engineering blog: Creating in-house source control. In May, we merged around 2,500 code changes per day. In September, we merged more than 30,000 per day. As a result, we became bottlenecked by developer infrastructure built for human (rather than agent) scale. Here, we share how we built a new version control system—and fully migrated our monorepo from git—in five months: anthropic.com/engineering/re…
5
222
I need an "Accept (but still taking other offers)" response for meeting invites
3
58
Muesli is indeed awesome. On-device speech-to-text with the best models and super-fast response times. If you write a lot or use only macOS built-in dictation, give it a try.
1
2
5
215
No one expected the way that AI ended up wiping out humans: The Funniest Joke in the World
JUST IN: Former Anthropic employee reveals the company has a version of Claude specifically trained to “solve humor.”
5
154
You think you’ve got a reasonably large English vocabulary, then you read some early Cormac McCarthy and realize, oh. English is just a monster of a language, ain’t it.
1
88
Michael Pipolo retweeted
this meeting could've been a decision by one competent individual
13
96
2
1,026
16,606
qualia fetishist comes to his senses
43
356
12
3,185
55,953
Digger: substantially nuttier and more ambitious than the marketing let on. Iñárritu mighta gotten a little out over his skis, but how could Warner have thought this would turn a profit?? Good movie though.
1
7
216
Michael Pipolo retweeted
The absolute evil genius of the safety team at Anthropic. You want to stop open the Chinese government from releasing open weight models? You could ask nicely. Or you could whisper the words “abliterated GLM 5.3” where the public security community can hear it.
8
8
2
271
17,749
oh shit now we're screwed
🤖 Made with AI
2
79