apothegm

26 ms·

481.
by apothegm 10mo ago
The issue is how terrible the LLM is at determining which sources are relevant. Whereas a somewhat informed human can be excellent at it. And unfortunately, the way search engines work these days, a more specific search query is often unabl
482.
by apothegm 11mo ago
I find it very hard to believe anyone who claims their AI tool “doesn’t hallucinate”. And how would on accomplish that?
483.
by apothegm 11mo ago
Doesn’t mean the author didn’t offload the task of writing to an AI. Also, Forbes blogs don’t have editorial oversight, only the official magazine articles.
484.
by apothegm 11mo ago
Good article, and easy to follow. I learned more than I’d expected from this one.
485.
by apothegm 11mo ago
Well, that was a thoroughly incoherent ramble.
486.
by apothegm 11mo ago
What do you mean by “need attention”? What problem are you actually having?
487.
by apothegm 11mo ago
Most stakeholders only need to hear about major milestones. They don’t get release notes. They get weekly or monthly memos on what the department is accomplishing. Unless you’re developing open source or developer APIs/SDKs, end users
488.
by apothegm 11mo ago
A lot of the ways we produce food are more environmentally damaging than they have to be. Runoff from farms that are careless about containing it and the use of finite aquifer water to grow water-intensive crops in desert regions being jus
489.
by apothegm 11mo ago
And we’ve all been regretting it ever since.
490.
by apothegm 11mo ago
If you don’t know the language better than the LLM, you can’t notice when it’s making terrible decisions. Use the LLM to accelerate your ramp-up and make you more productive while learning, but it’s not a replacement for learning.
491.
by apothegm 11mo ago
The most obvious value here is for HTMX, which requires a lot of partial templates.
492.
by apothegm 11mo ago
Depends. Who’s your user base?
493.
by apothegm 11mo ago
Not to mention that it’s obviously ridiculous to do things for reasons other than economic or financial.
494.
by apothegm 11mo ago
You’re a student. Open sourcing the code for portfolio purposes is a good idea, whether or not the code is great. Entry level engineers on their own are not expected to produce flawless code let alone architecture. That you achieved somethi
495.
by apothegm 11mo ago
The point of an exercise like that isn’t to have the code that walks a directory. It’s to learn how things are done in Python by writing that code yourself. Most people retain a lot more if they have to piece together the code themselves th
496.
by apothegm 11mo ago
No. LLMs are NLP engines. LLM-based chatbots can be used as mediocre, hallucination-prone search engines if you so choose. They’re especially bad if the answer you’re looking for is not an average of “collective wisdom” or opinion but a ver
497.
by apothegm 11mo ago
Did anyone besides Zuck himself not see this coming from the start?
498.
by apothegm 11mo ago
This happens to me in the US if I use certain DNS servers to connect. Have you tried switching up your DNS servers?
499.
by apothegm 11mo ago
Population size is usually included in those calculations. It’s typically GDP per capita . But I couldn’t agree more that the inequality and social safety net (or lack thereof) make the numbers deeply disconnected from QoL. Which I believe
500.
by apothegm 11mo ago
Even assuming it’s true, the fallacy in the quote is ignoring all the unreasonable people who try to adapt the world to themselves in ways that are harmful to everyone else. If the first sentence is true, then all destruction also depends o
501.
by apothegm 11mo ago
LLMs have basically solved most NLP problems. It’s the one thing they’re consistently good at.
502.
by apothegm 1y ago
It might just be an unintentional side effect of an older rule. They don’t want people running full-on nested VMs probably due to security concerns (inability to properly lock down the nested VMs). They wrote that rule in, say, 2005, and ha
503.
by apothegm 1y ago
Quantify the opportunity cost. How much money are you spending or losing by supporting a specific platform? How much revenue do you gain from that platform. Turn it into math.
504.
by apothegm 1y ago
Why couldn’t someone else make their own signing system, though, and apply it to arbitrary images?
505.
by apothegm 1y ago
And what prevents a malicious actor from doing exactly the same thing with an AI-generated image?
506.
by apothegm 1y ago
Have you tried BBEdit?
507.
by apothegm 1y ago
I mean, yes, duh, we should stop paying insurers. But I doubt the republicans are about to introduce Medicare For All, which is the only way to do that without vastly increasing the already unconscionable number of Americans who suffer or d
508.
by apothegm 1y ago
A few hours almost every evening that I’m home. Mostly pulpy fiction, though. I rarely have mental energy to spare by that time for books that require thought.
509.
by apothegm 1y ago
XMPP comes with the additional downside that most of the tooling around it is (or at least, a decade ago was) written in Erlang. Which is a niche purely functional language with a weird-ass syntax that is challenging to transition to from t
510.
by apothegm 1y ago
Totally agreed. It would be unreasonable to expect otherwise. That said, now that there’s more formal evidence of it, we’re in a better place to ask questions like “what are the systemic effects of that on policy?” and “are we okay with tho

More ›