falcor84

18 ms·

421.
by falcor84 8mo ago
Interesting, and my main take away is that ~16 million sessions is enough to distill Claude. That's extremely doable - obviously, as it's been done repeatedly - but it just looks very feasible in general. If I think of the number
422.
by falcor84 8mo ago
How did you get from reduced appetite to starvation state? I'm not on one of these drugs myself, but from friends I've been talking to, it seems like (after a bit of dosage tuning) it's reducing their appetite to what would b
423.
by falcor84 8mo ago
It's a nice post, just like the countless others saying the same thing over the decades. But it brings to my mind the infamous The POSIWID principle (the Purpose of a System is What it Does) - while readability is doubtless important,
424.
by falcor84 8mo ago
I'm sorry if you don't like pedantry, but this is what I'm in HN for. To be clear, I definitely didn't mean to imply that every coin in history is worth a thousand dollars, and suggesting that this is what I meant is cle
425.
by falcor84 8mo ago
A thousand years? What made you go with that number? The protestant reformation was only about 500 years ago, and I'm pretty sure that Martin Luther wouldn't have bothered that much if the expected "donations" were reall
426.
by falcor84 8mo ago
If you don't adopt biblical literalism, then isn't the Bible just true in the same way that Star Wars is true?
427.
by falcor84 8mo ago
I do know that, but I also know how donations can become an expectation. Also, it's worth noting in the context of this thread, that people can use AI inference for free on many services, with payment only need for higher usage, and ev
428.
by falcor84 8mo ago
If we look outside Christianity, what comes to mind is reading about the ultra-orthodox in Israel, and obviously about Iran.
429.
by falcor84 8mo ago
Where does the money come from? Religious services are generally funded by donations, and these donations are usually done in the open, whereby (from what I saw) regularly attending and not donating the expected amount would put you in a so
430.
by falcor84 8mo ago
Well ... isn't organized religion a subscription service for everyday life?
431.
by falcor84 8mo ago
While the two are closely related, I see a clear distinction between the two drives on their projection onto the explore-exploit axis
432.
by falcor84 8mo ago
> Vibe Coding has its uses and I'm sure that'll expand, but the idea of it replacing domain experts is outright laughable. I don't think that's the argument. The argument I'm seeing most is that most of us SWEs w
433.
by falcor84 8mo ago
Arguably it does, because being able to experience something gives you much more insight into whether it's right or not - so being able to iterate quickly many times, continuously updating your spec and definition of done should help y
434.
by falcor84 8mo ago
"The future is already here, it's just not evenly distributed"
435.
by falcor84 8mo ago
Yes, but you can do this at the top level, and then have AI agents do this themselves for all the low level tasks, which is then orders of magnitude faster than with human coordination.
436.
by falcor84 8mo ago
But there is a level of magnitude difference between coordinating AI agents and humans - the AIs are so much faster and more consistent than humans, that you can (as Steve Yegge [0] and Nicholas Carlini [1] showed) have them build a massive
437.
by falcor84 8mo ago
> AI will leverage me I think I know what you mean, and I do recall once seeing "this experience will leverage me" as indicating that something will be good for a person, but my first thought when seeing "x will leverage y
438.
by falcor84 8mo ago
Really, there's "no reason"? For me, watching ML gradually get better at every single benchmark thrown against it is quite a good reason. At this stage, the burden of proof is clearly on those who say it'll stop improvin
439.
by falcor84 8mo ago
And I'll just use this opportunity to recommend David Graeber's "Debt: The First 5,000 Years" - a wonderful book about the history of economy, demonstrating how it always boils down to effective use of debt. There are to
440.
by falcor84 8mo ago
But that's how ML works - as long as the output can be differentiated, we can utilize gradient descent to optimize the difference away. Eventually, the difference will be imperceptible. And of course that brings me back to my favorite
441.
by falcor84 8mo ago
How is that the takeaway? I agree that it's clearly they're not "alive", but if anything, my impression is that there definitely is a strong "semblance of consciousness", and we should be mindful of this sembla
442.
by falcor84 8mo ago
The immortal soul superpositioned in the neuronal microtubules.
443.
by falcor84 8mo ago
> As long as you have this architecture you’ll have this issues. Can you say more about why you believe this? To me, these questions seem to be exactly of the same sort of question's as on HLE [0], and we've been seeing massive
444.
by falcor84 8mo ago
Interesting - why wouldn't learning from simulations address the problem? To the best of my knowledge, it has helped in essentially every other domain.
445.
by falcor84 8mo ago
There's also the issue that whatwg pretty much stopped adding new semantic elements; if I'm not mistaken, only <search> and <dialog> were added in the last decade, after the introduction of HTML5. <main>, <sec
446.
AI-Assisted Software Architecture: Generating the C4 Model and Views from Code(workingsoftware.dev)
2 points by falcor84 8mo ago|0 comments
447.
by falcor84 8mo ago
I'm not a doctor, but am amazed that we've apparently reached the situation where we need to use these kinds of complex edge cases in order to hit the limit of the AI's capability; and this is with o1, released over a year a
448.
by falcor84 8mo ago
> Google has tried (a few times) to get chat to work The original gmail-integrated gchat/google-talk first released in 2005 was fabulous. If they had just kept developing it instead of repeatedly creating a new one, they would easil
449.
by falcor84 8mo ago
Brand New? It's over two years old now. For a company that brags about its cloud infrastructure and developer tooling, I find it incredibly ironic that Gemini has worse integration with Google docs than ChatGPT and Claude.
450.
by falcor84 8mo ago
Are you concerned that this will bankrupt Microsoft?

More ›