the_af

17 ms·

331.
by the_af 5mo ago
That's a very broad claim. Can you elaborate?
332.
by the_af 5mo ago
This is a hard experiment to conduct. I both agree with you that this is some form of "mechanistic"/"pattern matching" way of capturing of intent (which we cannot disregard, and therefore I agree with you LLMs can c
333.
by the_af 5mo ago
> Dogs we changed by accident (I know your point wasn't about dogs either, it just reminded me of something). I love Neil de Grasse Tyson's line in Cosmos: A Spacetime Odyssey : "This wolf has discovered what a branch o
334.
by the_af 5mo ago
> That is not the definition of a placebo. But, puzzlingly enough, it's the definition of open-label placebo , in which the patient is told they've been given a placebo. And some studies show there is a non-insignificant eff
335.
by the_af 5mo ago
I like the suggestion to emphasize the robotic/nonhuman nature of AI. Instead of making it sound friendlier and more human, it should by default behave very mechanistic and detached, to remind us it's not in fact a human or a comp
336.
by the_af 6mo ago
Nappies are no big deal but I honestly think it's simply some switch in the parents' brains that goes "let's not worry too much about this, it needs doing". You filter out the gross and you simply do it on autopilot
337.
by the_af 6mo ago
> I tried some ML language once, it's difficult even to write a basic factorial example What do you mean? It's one of the first things taught in any tutorial for the ML family or Haskell.
338.
by the_af 6mo ago
> If feedback from this incident is in its context window, it is highly unlikely to make this same mistake again. In my experience, this isn't true. At least with a version or so ago of ChatGPT, I could make it trip on custom word
339.
by the_af 6mo ago
Re: the confession. In my opinion it's meaningless. No LLM is capable of introspection; you cannot ask it why it did something, anything it replies is a "plausible conversation", not something it knows about its own behavior.
340.
by the_af 6mo ago
I mostly don't tweak anything at all these days. Stock Ubuntu, binary NVIDIA drivers, no tearing on video, excellent frame rate on most games. A few give me trouble (Space Marine II, one of the Need For Speeds). Sometimes the advice on
341.
by the_af 6mo ago
> Desktop Linux has always been kind of a joke. It hasn't gotten better, the problems are all still there. Desktop Linux mostly works these days. It does everything most regular people would want of it, with zero fuss. Including p
342.
by the_af 6mo ago
I love pixel art and specifically monochrome pixel art like this. It's a pity this blog was so short lived, I can only see 7 entries and only 2 Hokusai prints. Oh well, my own blogs usually don't fare much better.
343.
by the_af 6mo ago
Who knows what's true, but the official US narrative is that they entered his bunker, slaughtered the (mostly Cuban) security guards, and stopped Maduro just before he could hide behind a reinforced door. So the official narrative is i
344.
by the_af 6mo ago
Arguably the Soviets were on par (or more) than the US in defeating Nazi Germany. (Yes, lend lease, etc, this is not downplaying US contributions). The Eastern Front was the real battlefield were Nazi Germany was doomed. Stalingrad, Kursk,
345.
by the_af 6mo ago
"The Communist" were a faction in a civil war, that's not an invasion. And the split in both cases (Vietnam and Korea) was recent and artificial, in the sense of no tradition of there being two countries. It wasn't one c
346.
by the_af 6mo ago
> Vietnam and Korea were technically wars to stop conquest, no? No. For example, the US got involved in Vietnam to help the colonizer (France) stop an independence movement. Yes, because they feared the resulting Vietnam may become com
347.
by the_af 6mo ago
Agreed that arguably no strategy could have helped them against the Soviet Union, it was a major blunder going to war with them. But the Nazis self-sabotaged constantly. The Panzer IV and the Stug III (with the outdated Panzer III chassis)
348.
by the_af 6mo ago
That's not true. They could have standardized on a few rugged platforms -- and in fact, some in Nazi Germany advocated for that -- but their industry and engineering were generally self-sabotaging and a mess.
349.
by the_af 6mo ago
The thing I don't get about accounts of NDE and what people say about them afterwards is this: if they lived to tell the tale, their near death wasn't actual death. They didn't "peak over to the other side". So what
350.
by the_af 6mo ago
I do remember dreams. The times where I was anesthesized it was that on/off switch, I completely lost time. No NDE or even dreams.
351.
by the_af 6mo ago
> What will almost certainly happen is the same as every other time. This is what seems to me like a failure of imagination. As I said, I envision other possible and even likely futures. I'm not an oracle so I don't guarantee
352.
by the_af 6mo ago
> Doomerism is the easy path. It takes no imagination to repeat doomer memes and sci-fi dystopian tropes, without articulating exactly how we get there. It's not "doomerism" because there is a call to action, impractical
353.
by the_af 6mo ago
I'll add to this that 80% of any discussion of LLMs is instigated by CEOs of AI companies, and they themselves seem to believe scifi is a real-world education. So yes, it's a bunch of scifi-addled selfish amateurs guiding and pred
354.
by the_af 6mo ago
This is actually an argument I find convincing. The powerful need the less powerful to exist, because otherwise in relation to whom would they be powerful? Who would show them they are powerful? But even then, how many of the others would t
355.
by the_af 6mo ago
> So we reach this post scarcity society A full automation society, where the implied post scarcity is not necessarily for everyone. Maybe it needs most of the population not to exist in order for the few to enjoy the lack of scarcit
356.
by the_af 6mo ago
> If there is mass unemployment, who is going to buy anything from anyone? The "few" don't need or want us to be scraping in the dirt. > They want us spending lots of money on their products, so their wealth increase
357.
by the_af 6mo ago
It's not about capability. It's about who "holds the key". And sure, many currently with deep pockets and pushing for AI will miscalculate and get pushed by the wayside. I think many people who are not in the 0.001% are
358.
by the_af 6mo ago
That's not the rebuke you think it is. You made a claim (not original, I've read it before), someone expressed doubts about your claim (which if proven false, will have dire consequences) and you cannot wave it off with "ther
359.
by the_af 6mo ago
There are more options: Mass unemployment, consolidation of all AI-related benefits in the hands of a few, an increase in demand that doesn't outpaced the loss of employment, increase in capabilities (not AGI) that mean a few chosen pe
360.
by the_af 6mo ago
> It can't. It can't even deal with emails without randomly deleting your email folder [1]. Saying that it can make decisions and replace humans is akin of saying that random number generator can make decisions and can replace

More ›