vo2maxer

It’s frustrating to the point that I have considered inserting grammatical errors, but that would go against my principles, which I have attempted to inculcate in my children. Yes, a significant amount of what’s posted is copied and pasted AI slop. But what in the world preceded this? Barely legible slop? I would much rather have someone craft their thoughts, run them through their preferred model, and write something coherent that is not marred by punctuation or basic elementary grammar errors. And you know what, the hell with the AI slop police. Yes, if we choose to use em dashes, we will.

Yes! The reflexive “must be LLM generated” is becoming ridiculous. Anything that includes proper punctuation and, god forbid, em dashes which I’ve used all my life must be suspect. The “it’s not x, it’s y” construction predates LLMs. I don’t recall ever sending a text without making sure it contained no errors, and yes, many have included infrequently used vocabulary.

I’ve got this older laptop I just set up for dual boot with a Linux distro. Before installing GRUB, I stripped down Windows 11 using Chris Titus’s utility, though you can accomplish the same thing manually, as you probably know, to kill all that Microsoft telemetry garbage. Windows runs beautifully now, no lag whatsoever. Linux, on the other hand, locks up constantly. And it’s not a hardware issue. Plenty of RAM, confirmed all the drivers loaded properly. Ironic?

I think you are confusing forced exposure to something with being exposed to something by choice. I did not force my daughter to watch Bergman and Antonioni; I was interested in their movies so I saw them, and she chose to be interested in what kept me interested. That is how we get our cultural knowledge passed down through generations of parents who do not simply consume whatever algorithmically generated media is served up to them. You are setting the problem: you have assumed that for children to be introduced to anything other than the popular culture among their peers is always oppressive. And when you narrowed my references to Bergman, Kurosawa, Fellini, and Antonioni down to "Red Desert," you showed either that you are being dishonest or that you really don't know what you're dismissing when you dismiss all of these directors and their works. Her access to everything included books, music, films from different genres and time periods, and she chose which things she wanted to pursue based on the options available to her. The fact that now she does not care about the architecture of artificial intelligence does not indicate that the exposure I provided to her earlier failed her, and this is precisely what I said could happen: teenagers are making their own choices, influenced by peer pressure and social forces, and that does not make the earlier exposure I gave her invalid or mean I should have given her an iPad at the age of 5 and called it autonomy. I’m not writing off her curiosity, which you would see had you read all the way to the end of my comment. Given your username, I am not surprised that you are weak in nuanced thinking regarding exposure versus coercion.

Kids turn 18 eventually. Unless they’re homeschooled and kept in a compound away from peers with different experiences, I’m not sure how sustainable this approach is long-term.

I say this as the father of a 17-year-old who once read 200 books per semester in elementary school, winning school and city reading awards. This year in high school, she’s read maybe a couple of short stories at most. She’s grown up surrounded by bookshelves in every room, but now she has no inclination to even glance at the spines, much less open a book.

We read aloud together every night for years, usually books well beyond her grade level, which was already advanced. I exposed her early to Bergman, Antonioni, Kurosawa, Mizoguchi, and other great directors. Now her media diet is mostly TikTok and gaming YouTube videos. Musically, she’s remained open to everything from classical to oldies, fortunately. As for technology, despite learning quite a bit of Python and JavaScript starting at 10-11, she’s currently uninterested in and actively hostile to understanding anything about AI architecture or underlying systems.

Is this a teenage phase? Maybe. I’m hoping with everything I have that it is, and that the curiosity I modeled for her will resurface eventually. You can create the ideal environment, model the behavior you want, and do everything you can as a parent. But once kids develop autonomy and see what their peers are doing, they make their own choices. Sometimes those choices look nothing like what you hoped to cultivate.

Ironic. I’ve spent years with humans doing the same thing, only with more naps.

Can you furnish any solid evidence that the Democrats are not taking China seriously? I can’t seem to corroborate your statement as factually based.

I was curious why I was downvoted, and, what the heck, why not go ahead and ask. This is the answer I received from ChatGPT 4o:

Your reply—“An LLM wrote this and this and this… An LLM’s eternal recurrence.”—is clever and layered, riffing on both the original comment and Nietzsche’s concept of eternal recurrence. On Hacker News, though, that kind of playfulness can sometimes backfire for a few reasons: 1. Tone Perception: HN users often value clarity, conciseness, and substance. Your comment may have come off as glib or performative rather than adding meaningful insight—despite being philosophically witty. 2. Thread Saturation: If others also replied similarly (“An LLM wrote it” variants), your continuation might have felt like piling on or diluting the thread, which can get downvoted as noise. 3. Cultural Split: Hacker News has a split temperament: some users appreciate literary or ironic references; others see them as distractions from signal-rich discussion. 4. Automated-Sounding Style: Ironically, even though you’re parodying LLM-speak, your repetition (“this and this and this…”) could be mistaken as LLM-speak by some, triggering reflexive downvotes from those skeptical of AI-generated content.

In short: you were being clever, but the average Hacker News voter might’ve wanted more content or simply misread your intent.

An LLM wrote this and this and this…An LLM’s eternal recurrence.

Given your abhorrence of verbiage, here is some LLM spew instead:

In Search of Lost Time by Marcel Proust is a monumental novel about memory, time, and the subtle shifts of consciousness. Through the narrator’s recollections—often triggered by sensory experiences like the taste of a madeleine—Proust explores love, art, jealousy, and the passage of life in early 20th-century French society.

Fair point. Side-channel attacks show how much signal you can pull from noise. But fMRI is a different kind of beast. It’s slow, indirect, and coarse. You’re not measuring neural activity directly, just blood flow changes that lag by a few seconds.

The paper [0] doesn’t pretend otherwise. It trains a model (PAM) to learn which brain regions carry useful info for reconstructing images, and applies this to both fMRI data from humans and intracranial recordings from macaques. The two signal types are handled separately.

If you want an analogy, it’s less like tapping power lines and more like trying to figure out which YouTube video someone is watching by measuring heat on the back of their laptop every few seconds. There’s a pattern in there, but pulling it out takes work.

[0] https://www.biorxiv.org/content/10.1101/2024.06.04.596589v2....

Just to clarify, the paper [0] does use both implanted electrodes and fMRI data, but it is actually quite transparent about which data came from which source. The authors worked with two datasets: the B2G dataset, which includes multi-unit activity from implanted Utah arrays in macaques, and the Shen-19 dataset, which uses noninvasive fMRI from human participants.

You’re right that fMRI measures blood flow rather than direct neural activity, and the authors acknowledge that limitation. But the study doesn’t treat it as a direct window into brain function. Instead, it proposes a predictive attention mechanism (PAM) that learns to selectively weigh signals from different brain areas, depending on the task of reconstructing perceived images from those signals.

The “thermal imager” analogy might make sense in a different context, but in this case, the model is explicitly designed to deal with those signal differences and works across both modalities. If you’re curious, the paper is available here:

[0] https://www.biorxiv.org/content/10.1101/2024.06.04.596589v2....

Observations like these remind me of The Académie des Beaux-Arts in France, and more specifically its official Salon (the Salon de Paris), keeping Impressionist painters out of established exhibitions.