Hacker News Reader

Comment by infwhispers | Hacker News Reader

I am the author of the research. Please, feel free to ask me anything. Also, check out the followup research: https://8rank.ai/game/apple-intelligence-knowledge-graph-in-...

View on HN

Please stop using AI to author your articles. It was really hard to read it despite how interesting the topic was. I dropped halfway because it was just too annoying.

My bad. In my defense, only the technical parts are AI. But, thanks for the feedback; I thought I had unsloped it enough...

> In my defense, only the technical parts are AI.

A defense would be "I just used it for spellchecking" not "I just used it for the meat and the important parts people care about, the rest is mine!"

Yes, that was a weird defense; essentially, he used it for the actual interesting parts of the article.

So most of it?

So only the parts that matter, you ai slopped. Great work.

You can’t unslop A.I. prose. Once it’s written, the content is poisoned.

“Sure there was a turd in the punch bowl but I did my best to remove it, so you should be able to drink it.”

Nope.

And the technical parts are what matter, saying you did that with A.I. tells me “I don’t care about this part” so what the heck are we even doing here?

Interesting for me it's the opposite. I care about this so I let AI do it since LLM are better writers than the majority of people. I'm not so arrogant to think I'm a writer of that caliber.

You are. I promise you. Almost every single person is a better writer than an LLM. You're wilfully sabotaging yourself.

LLM are terrible writers though. Worse than most humans unless they have some mental disability.

Maybe LLMs are better writers than the majority of people, but that's only because the majority of people just don't write. If you have any level of competency, you are a better writer than an LLM, because good writing requires empathy. It required thinking about what the audience knows, what they want to know, what kind of writing they are comfortable with, and so on.

LLMs fail at those things.

It's not possible to un-AI text, only replace it wholesale.

Suggestion for next time: "Review my prose for accuracy, grammar, appropriate strength of claims, etc. Flag any issues for my attention. Do not provide text; this work should not contain any AI-written content." That'll turn it into a very good check that doesn't take your voice away.

I second this. Unfortunately you just can’t just “unslop” it, much like you can’t just “shake the dirt away” from food that fell into a bucket of… you get my point, it won’t be good to eat anyway. Bonus additional ironic parallel could magically manifest if you can guess the bucket content.

Anyway, it’s a shame, this could have been excellent work. Instead, I find reading it unpleasant and part of my attention rots away wondering if you even checked what the LLM wrote for you. You probably did, and I could also do it myself, but still…

I was bothered not so much by the slopful article as by the AI-written text in the dashboard that iKnow produces. I ran the program on my Mac and it did display charts and graphs with names of people and organizations I’ve interacted with over the years, but I couldn’t figure out what it all was supposed to mean.

One example: "323 people, 122 events and other records, 200 employers and 1 stored group — as the store on this Mac links them. Read live just now. Colour is what each node is to you: the stored group in red, family and closeness by iKnow’s rules on the kinship scorecard. Size is how many rows in the store name it."

Claude etc. can be prompted to produce clearer explanations than that.

Yeah seriously. Article becomes unreadable immediately after the intro. I believe this author probably could actually write two years ago, it's a shame people are just giving up on that and choose to turn out stuff like this instead.

I'm having more trouble spotting AI text these days. It used to jump out at me. I'm not sure if it's that the LLMs are better or it's just becoming so normal to see that I stop recognizing it. This one didn't scream "slop" to me other than being quite long-winded for a blog post that I admittedly did not finish on first read. Going back and scrolling deeper into the post, it does become much sloppier.

> quite long-winded for a blog post

That's actually exactly what does it for me. After the intro, it gives you way too much specific detail. No one cares about the exact file path or whether that path is TCC Full Disk Access protected. What does that even mean? Why is this important enough to be in the first sentence of the section? I don't care that there's 51 database files and 181 other things and 278 doodads. Who says stuff like "triple store?" It uses bizarre language with far too much detail, and like you said, your eyes start to glaze over because there's so much irrelevant information. No human would've chosen to write all that tedious detail in their post about something much more interesting. It's bad writing. It's slop.

A human would've said something like "It's a database in <Library/IntelligencePlatform> with three columns that contain..." and left all the rest of it out or maybe in some extreme details footnote.

I disagree. Hand-authored articles can be slop too. This research is great, and your subjective feelings about word cadence don't affect the information. I guarantee you read AI output every day and don't know it.

The criticism can be restated, then, as “you used AI to write it, and unlike others who use AI to write things, you did a bad job directing it and editing the work.”

This comment is not directed at you, but this is the best place I could find to insert my thoughts on this "nO LLm slOP" distraction:

----

The research is jaw-dropping enough for the average reader to comprehend (e.g: I am not a techie, but a former IBEW datacenter wiretwister).

It could be covered in pube-sprinkled-turds and would still be worth reading.

To me the "slopLOL" crew is akin to people who'll take their time to correct others' their/they're/there usage:

Just read the fucking sentence (LLMs can and do and aren't so poopy-you-suck-for-being-human).

We know there’s interesting information in there. That’s not the point.

How news is delivered matters. If you don’t believe that, why would news outlets compete with each other? Even if the facts are the same, would you prefer to get them from a good or bad writer?

>would you prefer to get them from a good or bad writer?

I am smart enough to not need perfection. I just want the data. Write however tf you want, good'll'be read.

And so will LLMs.

I'll suggest adaptation.

>why would news outlets compete with each other?

Competition for the masses' con$umption$?

It seems to generally work; shouldn't have to obfuscate the `man` pages for this always-running daemon... how many people even know what `man`pages are (obviously not here)? how to access them? why they would even want to?

Well, good for you for not caring, I suppose. The vast majority of society does.

> I am smart enough

Knock yourself down a peg, kid.

I will, sure, but I've been writing about this for almost a decade; and I know more than most people about this topic, despite being a bluecollar electrician.

Also, my IQ scares even me; I've chosen bluecollar so I don't have to be evil as much with it.

----

Companies do not give two shits about most users, with (e.g:) AMD saying the quiet part out loud, recently, with their recent gamers-last announcement.

Also see: Apple's invasive attempts to categorize all local data, by default; if you want to opt-out (some service, like OP's, you cannot!), it takes dozens of clicks for each new software – intentional dark-patterns to make opting-out cat-and-mouse.

----

>Well, good for you for not caring, I suppose.

I care too much, is the problem (at least from my POV). Companies are soulless, and only care for shareholder value (legal requirement).

Data is the new_oil is the new_steel is the new_tin is the...

> I've been writing about this for almost a decade; and I know more than most people about this topic

> Also, my IQ scares even me

Oh, brother.

I mean this with the best of intentions and care: have you been psychiatrically evaluated lately? The language you’re using is consistent with friends I know who are off their stabilizing medications.

Hand-authored engagement or clout (LinkedIn) farming articles and corporate press releases are usually slop, most others aren't. I'd be surprised if corporate press releases today weren't largely AI-written - no one would know the difference.

There was a very obvious and unnerving shift in style that happened early in the article where it went from human to LLM. I find such shifts really glaring. Agents, in their current state, simply cannot produce quality prose and are not good enough to write blog posts targeting a human audience.

I found the topic fascinating, but eventually I started skimming and then gave up on it.

Hi there. Thanks for this!

So my questions regarding security implications of running your code, my LLM friend reviewed the code and expressed possible concerns over (a) malicious browser extensions now get to see what they couldn't before; (b) you are not using https. There was bunch more but they can all be addressed on my end. Your thoughts?

You don’t need https for a web running locally; nothing goes into the network. Malicious browser extension can be real, but if that is true, you might have bigger problems than leaking the apple graph

It says there's 137 live/certain facts in my graph. I can see where they're sourced from/a statistical breakdown of type/the number of unique confidence values/how many citations they're using/how many ingests contributed no facts. Am I missing where I can see what the actual facts are?

A lot of these statistics aren't very useful for mining insights from the graph, from a technical standpoint of how the graph works or from a personal standpoint of how the graph is interacting with my personal data.

I'm on macOS 27.2 and there is no longer a "Full Disk Access" control. How do I use it in this case? ChatGPT is not helpful.

open "x-apple.systempreferences:com.apple.preference.security?Privacy_AllFiles" << this made it reappear in settings

The data I see from iKnow is a sloppy clusterfuck, I suspect from botched syncs and lazy bookkeeping on my part, and the fluctuating quality of Apple's "services" over the years. My Mac settings probably go back to the 1990s and Apple periodically did a bad job. Pain points I remember: Google contacts being separated and then with a new phone not syncing correctly, Mac address book burping on contact avatar images, importing lists of contacts from short-lived 3rd party apps, shared Photos fucking shit up, the iPhotos upgrade fiascos, transient contact info for my kids, moving onto/off of Gmail, etc.

If the FBI got ahold of my (and others'?) iKnow data, a lot of people might end up in a sloppy jail with a sloppy judge sloppily applying a sloppy law...

Thanks for the research. Have you looked into how to secure / disable the subsystem?

(I hate that a discussion on the OP has been hijacked by a discussion on AI writing; the substantive questions are all at the bottom. I wish mods would 'detatch and move' the AI subthread.)

You wrote that other apps can't access it because of TCC Full Disk Access. Presumably the OS itself can read it, so do you have any information on whether parts of the graph (or fingerprints, or inferences made with it) are ever sent to the cloud/apple servers?

Reply on news.ycombinator.com