Currently reading Nexus and feeling increasingly bleak about the future...

Funny small world story. I actually walked across a Ferrari today. Yeah, the proper idiom is "ran across", but I was walking at the time. (Can't really recall if I've ever seen one before that wasn't in a dealer's showroom, though I recently spotted a second BYD in the wild.)

Out on my daily walk when I noticed an unusual looking car by the side of the street ahead, so I crossed the street to get a better look. When I got there it turned out to be a Ferrari. And the proud new owner was telling a couple of friends about the car.

Crazy and funny. No place around here where such a car can be driven at any significant speed. By here I mean the entire country.

Disclaimer needed? More than half my life since I stopped driving. Only ride in a car a few times a year, and some years get skipped.

If a real expert did the work of validating the bug reports before submitting them, then I don't see the problem here. At least for the time being. The AI would have served as a useful tool, even though the bugs need to and should be validated a couple of times. Just part of the process of fixing the software without introducing new bugs or regression bugs. (But of course the worst bugs are going to involve interfaces with other software...)

So from that relatively optimistic perspective, you could argue this is a new category of job for human beings: Managing AI in various ways. Writing the right prompts. Asking the crucial questions. Testing for AI slop and hallucinations.

At least until better AIs reach that level.

However I see the biggest problem is the production of future experts as in NOT. I don't know anything about this Joshua Rogers, but I'd wager that he worked hard and long to become a security expert of any sort. Projecting from my own experiences, but figuring out which level to attack from and tracking what is referencing what and figuring out which black boxes need to be opened and figuring out which specifications are broken... All those sorts of things are hard work, but AIs are getting in the middle now and if that work goes away, then so will the learning from experience.

[So much for my initial reaction. Now I have some keywords to search for, since the moderators are so useless. And I'll check for Funny, too. Another place where moderation has become increasingly useless over the years. Hmm... Maybe Slashdot could fix the moderation system by using AI modbots?]

Not anything. I'm sure when Federal troops take over the State Capitol and Newsom is put in prison for unspecified but certainly horrible crimes, the military governor that takes his place will make sure none of this kind anti-corporate nonsense continues.

I'm having trouble interpreting your reply, but it sounds like you are advocating for active forms of genetic intervention whereas I have fairly strong reservations against such intrusions. Rather I think the right to reproduce should be guaranteed, but we can legitimately look at how the dice landed before an actual human being is involved... Ma Nature's evolutionary approach is basically blind and the bad shuffles merely become food and fertilizer faster.

But I may be an evolutionary extremist. For example, I can imagine keeping frozen embryos as a kind of reproductive insurance for cases where your child doesn't live to reproductive age... Plus special insurance for new mutations provided at a governmental level.

My main reaction to this article was "Yet another reason not to use Firefox on my smartphone". (Main reasons are privacy and security concerns. I prefer to keep some eggs in separate baskets. Another reason is resource demands on the smaller device.)

However your FP triggered a different question, though I also feel a bit rude to ask you for clarification and thereby exploit your hard reading of the full story beyond the Slashdot summary. Nothing above you about the google, so...

Based mostly on your FP, I think the full story here is that Firefox took money to work the google's Gemini more deeply into Firefox. I certainly understand the google's motivation even though so many of the Gemini results cannot be understood. And I even understand that Firefox's motive is clear enough given how broken their business model is.

However I always approach such "new feature" stories from the perspective of "Would I pay for that?" And the answer is almost always negative. I'm not saying that I could pay as much as the google, but if a LOT of people agree with me and we all chipped in, then maybe we could have more of the features we want and fewer of the features we hate. Firefox has implemented a number of features that I would pay to go away...

Too bad we don't live in such a world, eh? We just have to suffer slings and arrows of outrageous features that we have to adapt to. Okay, I do like learning new things, but not so much when they are bad new things being rammed down my browser's throat.

On the one hand, the negative moderation and your comment make me wonder if that was a situation where anonymity was justified.

On the other hand, I'm wondering if his use of anonymity encouraged his post to go overboard?

Me? I was looking for references to genetic counseling, but you might be in favor of dead children. The way Ma Nature sees things involves equilibrium, and the equilibrium calls for more dead babies in the sense of having kids who don't reproduce. The math is actually quite simple. The genes are shuffled at random, so half the children are luckier than average and half aren't. The "natural' solution is four or more kids for each couple, but only two of them live long enough to reproduce. And yet parents tend to dislike that approach. (Threatening to go overboard, but I would argue babies are so helpless that our love for them has to be in some sense excessive or absolutely none of them would survive.)

Kind of hard to interpret your FP, but that's how "the rush to FP" works, isn't it? The "it" in your Subject must mean "gibberish" or "commit the act of spewing gibberish", but I'm focusing on the "reason" part as an implicit question about why. On that basis I react negatively to your simplistic and absolutely dismissive answer. And I offer my Subject as an answer to your implicit question.

I do sort of agree with your negative sentiments towards LLMs, but I think they cannot be dismissed so easily for two main reasons. One is that so many people are eager to find an oracle and they eagerly want to believe whatever their favorite genAI says, up to the point of rationalizing hallucinatory gibberish. Never forget that "People believe what they want to believe." (I think the first time I heard that was from Dr Bill Martin about 50 years ago, but it's still true and even seems to be getting worse.)

The second reason is that the blending effect of the massive input tends to make the LLM output sound plausible just because of its similarity to "the center of mass of" the training data. Whatever the genAI says, it always sounds like something you've heard before. Or more likely like many things you've heard many times.

But my reaction was that the specific problem of this story is more like a brainworm. Usually that's a song or musical phrase that gets stuck in your head, but it can be an image or idea that becomes a kind of fixation. Therefore I am quite interested in how these brainworms are triggered because the triggers might work on humans, too. If so, I'm sure the applied psychologists are already hard at work figuring out how to abuse us with that new technique. Least malevolently for selling us a different brand of toothpaste but at the worst levels pushing stuff like building support for civil wars. Can you imagine "We've got to sink the Bismark" mutated to "We've got to sink Chicago"? (Did I just trigger a brainworm? Long time since I've heard that song...) [And then I searched this discussion for any mention of brainworms, but there was only a kind of reverse reference in a sig...]

By the way, I also think this topic is related to how facial recognition can be poisoned with a recognition icon. Some years ago they discovered how to make a small image that when included in a larger image would trigger recognition of a particular person without regard to the rest of the image. My memory is already fuzzy, but they essentially distilled the signature of a person and then included that signature in any input image to create a high probability match without regard to the actual faces that actually appeared in the image. And now I just thought of a funny game to play with the technique. Put one on your forehead, a different one on your right cheek and a third on your left cheek and you become three people depending on which angle they look at your face. Scatter them about and you could become a multitude?

I think the international aspect is actually more significant. Turns out the third winner of the Chemistry prize is another one of those annoying immigrants...

Now you have me confused. That's the second Japanese-involved Nobel this year. The first was in Medicine, while the second is in Chemistry...

*sigh*

FP haste makes waste.

s/they I admit/though I admit/

Kind of a sincere question they I admit that I have a lot of theories about the answers...

Tease my biggest? Inability of the police to understand what's going on? Or maybe something about innovator's advantage to the crooks?

Related reading: How about Nexus by Harari?

Sounds like you're feeding a troll, but I don't even know what kind of food "gammon" is. But does it come in kosher and halal versions?

File "feeding troll" under "That trick never works".