roughly

18 ms·

391.
by roughly 10mo ago
> The fact that the business started to make money may have been in spite of the CEO, rather than because of it. One begins to understand why the C-suites are so convinced this technology is ready for prime time - it can’t do _my_ job, b
392.
by roughly 10mo ago
I actually enjoy some of the wonkishness. Hearing a good interview with people who are experts in a particular field is a great way to learn.
393.
by roughly 10mo ago
Yeah, but where do I keep that one?
394.
by roughly 10mo ago
I’ve been a fan of The Ezra Klein Show for years - whatever your thoughts of his personal politics, he’s a fantastic interviewer, and I think he does a great job at both helping define and then interrogating the strongest construction of hi
395.
by roughly 10mo ago
The ol’ TMA problem.
396.
by roughly 10mo ago
Old joke: A pair of mathematicians are sitting in a park chatting. There’s a snack shack near by, and while they’re talking, three people walk into the building. Some time later, four people walk out, and another couple walks in. A bit afte
397.
by roughly 10mo ago
Presuming that’s why they raised 3 times this year.
398.
by roughly 10mo ago
No path to profitability for the people using their products for their putative purpose, which seems like it might affect Nvidia’s bottom line at some point. Clarified.
399.
by roughly 10mo ago
The absolute best case I can make for this: I think it’s pretty obvious at this point that Nvidia’s architecture has reached scaling limits - the power demands of their latest chips has Microsoft investing in nuclear fusion. Similar to Inte
400.
by roughly 10mo ago
The Brookings institute is hardly a lefty rag - they're about as centrist/neoliberal capitalist as an institution comes. > Ignore what's in front of you Yes, the general advice is to look past specific notable anecdotes an
401.
by roughly 10mo ago
Reminds me of the old SNL sketch: https://streamable.com/m7omz
402.
by roughly 10mo ago
That’s a hell of a take for shoplifting.
403.
by roughly 10mo ago
> A few well known AI scientists believe that what happened with Transformers can happen again, and better, following different paths, and started to create teams, companies to investigate alternatives to Transformers and models with exp
404.
by roughly 10mo ago
I think it’s underappreciated the degree to which police and LEO have started behaving like political actors. NYC cops decided to “strike on the job” when the city started changing its stance in the mid-2010s, and in the Bay the cops respon
405.
by roughly 10mo ago
So the problem here is the banks wrote their loans stupidly and both the bank and the property owner's willing and able to burn money to keep those chickens from coming home to roost, which has an enormous social cost on small business
406.
by roughly 10mo ago
One of the things you get an intuition for after using these systems is when to start a new conversation, and the basic rule of thumb is “always.” Use a conversation for one and only one task or question, and then start a new one. For longe
407.
by roughly 10mo ago
No, but I’m also not an accessibility expert, so my opinion here’s pretty irrelevant.
408.
by roughly 10mo ago
> And I do not buy the argument that Calibri is somehow more accessible for those with low vision or reading disabilities Oh well that settles it, John Gruber doesn’t buy the argument. Wrap it up and let’s head home, folks, this one’s se
409.
by roughly 10mo ago
I think it's two aspects - one is that Insta started out as the "not-facebook" platform and Meta's somehow managed not to fuck that up, and the other is there's a massive network effect - every tattoo artist, venue,
410.
by roughly 10mo ago
This is a bit of a harsh take. These same groups use Signal for all their internal messaging - by and large the will is there, but they're not tech savvy and we haven't given them viable alternatives.
411.
by roughly 10mo ago
I've been working to get more involved with, eg, mutual aid groups and other forms of local capacity and resiliency building over the last year - one thing that's stuck out to me is how many of these groups' public face is an
412.
by roughly 10mo ago
I actually like the device and can see the purpose of it - I’ve tried to kick together similar solutions in the past, but I like the very “one thing only” nature of this. That said, I absolutely cannot buy a device like this without a repla
413.
by roughly 10mo ago
I know this isn’t Just a Samsung problem, but it’s sure not Not a Samsung problem either. Samsung’s taken the early-00’s Sony crown of most anti-consumer company out there - don’t buy Samsung.
414.
by roughly 10mo ago
Yeah, there’s a lot of critiques of the product/packaging/etc, but this feels like huge validation of the Framework model - this is an unrelated 3rd party looking to get a chip in consumer hands who decided to use the FW chassis.
415.
by roughly 11mo ago
Here’s the strongest steel man I can provide: A person typically picks stocks based on some view of the future: what’s going to be valuable, how the world is going to change, etc. Presumably, a congressperson does the same when drafting leg
416.
by roughly 11mo ago
More to the point, most humans loudly declare they would do the right thing, so all the chatbot’s training data is on people doing the right thing. There’s comparatively fewer loud public pronunciations of personal cowardice, so if the bot’
417.
by roughly 11mo ago
> I imagine using their embeddings and training a classifier on top of that is probably a lot more effective? I’d be interested in seeing this in action - I think the vector embeddings are underused generally - but my understanding is th
418.
by roughly 11mo ago
Yeah, my mental model at this point is there’s two components to building a system: writing the code and understanding the system. When you’re the one writing the code, you get the understanding at the same time. When you’re not, you still
419.
by roughly 11mo ago
I’ve seen a couple solo efforts and a couple teams, but usually a few months. It tends to evolve as a kind of whack-a-mole situation - “we solved that failure case/hallucination, now we’re getting this one.”
420.
by roughly 11mo ago
One of the things that makes it very difficult to have reasonable conversations about what you can do with LLMs is the effort-to-outcome curve is basically exponential - with almost no effort, you can get 70% of the way there. This looks am

More ›