I like OCaml. Formerly Microsoft (O365 Core/Substrate internals), Facebook (Hacklang). I identify as a peacenik
Joined August 2020
- Tweets 5,524
- Following 561
- Followers 423
- Likes 3,864
387 Photos and videos
The generation after Omega will be drawn from the Egyptian hieroglyphs, hopefully
16
I wonder if they (a hospital system in my area) prepared this message before they transitioned, and if they didn't, then why not? Premortem always a good idea
31
I have this particular musical blindness which is especially obvious when it comes to jazz. Unless there's harmonic movement, I don't perceive the music to be distinctive unless it's really really obvious
2
2
73
So for example, I'm very cold towards a lot of solo material. A lot of jazz music sounds the same/generic to me. My husband and my jazz-playing 16yo can usually identify a standard or a song they've heard before from a couple of bars, but they sound absolutely generic to me
2
2
62
Take, for example, Pee Wee Ellis's The Chicken. My family would probably identify it from the initial riff, but it doesn't sound like anything to me until the harmony starts to move, which is when I recognize it. And the solos are just blah, don't care
1
1
51
"Amazon called administration officials Thursday night to share a report showing how they were able to jailbreak and access portions of Anthropic's powerful new Mythos model" - wait, Anthropic seriously thought their stupid guardrails were bullet-proof?
1
55
I agree with the WH that Anthropic is unserious, but this is also performative bullying
axios.com/2026/06/12/anthrop…
31
Jun 12
Didn’t look into AI and EA connection before. First impression is it’s just like how hippies became ruthless and cynical businesspeople in the 70s and 80s. Again recalling How Music Got Free and the story (among many in the book) of Doug Morris, a music executive
1
30
Jun 11
And at this, I wonder if AI productivity and reliability fall under the No True Scotsman fallacy. It’s like, if 99.999% of people can make a good viz with Power BI, it’s useless to blame it on “skill issues.” Skill issues or no skill issues, the tech is not fit for purpose
Jun 11
Jokes aside, I tend to agree. In most cases where LLMs appear do something annoying or bad, it’s usually the harness, the orchestrator, the filters, the prompts - the parts bolted on my humans
1
1
129
Jun 11
But I don’t want this to sound like I’m saying it’s possible for PowerBI to be very good. It’s not
1
1
27
Jun 11
But it’s possible that some of the components that Power BI is put together from are very good. Maybe not for building data visualizations, but for something else
12
Jun 11
Claude vs GPT is Pepsi vs Coca-Cola. At some point, you just win the battle for consumers’ minds by proper positioning
1
52
Jun 11
My family does call me MomGPT ‘cause I do know some things and can search for things still. When the last of the millennials dies, that will become a lost art
Jun 11
Did it find it on its own, or did it read my reply that I posted a minute earlier? 😂
32
Jun 11
The current crop of AI-based tools is a form of automation between documentation and scripts. Docs is the first level of automation (think pilot checklists). AI should be used to take automation all the way to scripts and software. I.e., agents don’t replace SaaS, Mr. Nadella
1
64
Jun 11
Agentic use should be observed to extract patterns into workflows, pipelines, and other familiar forms of software that may only use an LLM for a small portion of the work, if at all. Analogous to software moving into a more efficient implementation in hardware. It’s a continuum
51
Jun 11
It’s not impossible to measure improved productivity without relying on layoffs AI replacing people to cut costs. But it’s more work and riskier. The old pitch was simple: Companies have decades of outsourcing experience, and AI is cheaper than outsourcing. If true, it’s low risk
Jun 11
Say, I have 10K employees, and I lay off 5% (500). If employees cost me 500K, then I can pay
500 * 500000 / 9500 / 12
So, just under $2200 per month per employee would be, cost wise, where I was before the layoff. So it has to be a lot less than that to make sense.
69
Jun 11
Are they ignorant or deceitful? LLMs don’t make the transcripts structured, searchable and queryable. We’ve been doing this for years. An LLM is the UI and the consumer of the transcript, which is not indexed into an LLM, obviously, nor structured by one.
Most work conversations are now being recorded by default. You should probably assume that everything you say at work is getting recorded from here on out.
What’s emerging is a new category of enterprise software, organized around voice instead of text.
The system of record today is structured data: CRM entries, tickets, docs. But the highest-value context lives in conversation: the nuance on a customer call, the real argument in a product review, the offhand comment in a leadership meeting that quietly changes the roadmap.
LLMs are uniquely good at taking that unstructured voice data and making it structured, searchable, and queryable. That’s a large enterprise opportunity, and we’re still early in understanding what the software layer looks like and who owns it.
a16z GP David Haber on what AI recording means for the future of work: a16z.news/p/everything-is-re…
2
54
Jun 10
If you wanted to spy on your own citizens, which professions would you draw your cadre of spies from?
1
99