5,519 karma · joined December 10, 2022
Email: raptors_03pout@icloud.com
They do though! They mostly try to avoid it since hitting the network to serve any kind of latency would unacceptably increase latency, but you wildly underestimated the amount of complexity there is to running a CDN.
Apart from the issue of having to compile the downloaded software leading to spinning fans, I'm not too concerned.
If LLM writing improves to the point where they can infer the business (or other real world) context and serves the functional purpose of informing others as opposed to being an intellectually lazy piece being produced only for the purpose of being produced, then I’d be fine. It’s likely the author would need to spend some effort on the said piece of writing, regardless of how good the LLMs get good at writing.
The thing that gives it all away is that they claim that the IP addresses are from Azure, and then proceeded to redact the IP addresses, as if they belong to individual users. It's laughable.
The IP addresses are the most interesting part of this experiment, as it would have provided researchers a way to understand the distribution of IP addresses used for the spam operation within the ASN.
* Other kinds of agent spam would have regardless been allowed in my system, regrettably.
It seems that the user wanted to say something like "DeltaDB is a DAG" and then replaced the DAG part with some obscure graph-based project that the LLM had suggested, but otherwise has no connection to agentic development.
I disagree with the "experts" there, as if latent-space reasoning will only cause proper interpretability research rather than taking the CoT as gospel[2].
Give me that any day over the constant "you will be abolished to the permanent underclass!" talk coming from billionaires.
Edit: (A better discussion is apparently here[3])
[1] https://arxiv.org/pdf/2412.06769
[2] https://thezvi.substack.com/p/the-most-forbidden-technique
1. For a given company, analyze their target audiences and the questions they are likely to ask LLMs about.
2. For each such question, ask it to each of the major LLMs, and compute the KL divergence between the pages they want to rank for the question vs. the LLM's response.
3. Rewrite the article to minimize said KL divergence.
In effect, they're performing an iterative optimization of some sort that moves the embedding space of their article closer to the question asked to the LLM, and any embedding model or generated responses are going to prefer said responses over others.
I believe we will keep seeing more of this stuff.
> People don’t like nitpickers. “He literally did the WELL AKTUALLY!” If you say Joe Criminal committed ten murders and five rapes, and I object that it was actually only six murders and two rapes, then why am I “defending” Joe Criminal?
> Because if it’s worth your time to lie, it’s worth my time to correct it.
> If one side lies to make all of their arguments sound 5% stronger, then over long enough it adds up. Unless they want to be left behind, the other side has to make all of their arguments 5% stronger too. Then there’s a new baseline - why not 10%? Why not 20%? This mechanism might sound theoretical when I describe it this way, but go to any space where corrections are discouraged, and you will see exactly this.
[1] https://krebsonsecurity.com/2018/03/who-and-what-is-coinhive...
I'm required to use Ubuntu at work. Coming from Mac, apart from the menu bar at the top, on Gnome, I've been able to customize the keyboard shortcuts, remap the keyboard so that Ctrl works like Cmd, and use extensions like Dash to Dock[1] and themes like Whitesur[2] to replicate something that almost works like a Mac.
The keyboard remapping and customizing keyboard shortcuts were all done within default the default settings app.
The only things missing are some keyboard shortcuts like Ctrl+A/V to move to the beginning and end, and the Ctrl+Shift+C/V behavior on the terminal instead of Cmd+C, which I've just worked around by using VSCode's terminal and configuring it to copy when I press Ctrl+C with some text selected.
[1] https://extensions.gnome.org/extension/307/dash-to-dock/
The worst one was where I fixed a datetime bug and although it had been sending out false alerts, I was asked to dry run the 5 lines of code I changed, like a coding interview. In all this pressure I forgot what the code was meant to do, and was dismissed and asked to set up another meeting with an explanation of all the various cases that could happen...
It's also a bit rich especially to complain when Microsoft also suffers from forced autoupdates and the like.
In effect, I’ve always wanted a pair programmer agent, not a zero to one programming agent. Unfortunately models these days are mostly of the latter kind and it has caused a major disruption in the way I work. I’d much rather appreciate a small model making fast and specific edits that I ask if it, rather than ingesting 20 files to make changes, and then starting to write tests, etc.
I don't think the current crop of fullstack engineers would be hard pressed to know what a "markov chain" is, but in theory yes, you could emit a bunch of speculation rules[1] based on your predictions.
I should also say that markov chain based approaches have been used for fraud detection, e.g. identifying checkout anomalies by detecting the sequence of web pages that they clicked on, amongst other factors.
[1] https://developer.mozilla.org/en-US/docs/Web/API/Speculation...
The other explanation may be that these AI labs may be expecting more government scrutiny, and "here's a document" would probably go better than "here's some vector representation of our values" when talking to politicians.
[1] https://arxiv.org/abs/2106.09685
[2] https://vgel.me/posts/representation-engineering/
[3] https://transformer-circuits.pub/2024/scaling-monosemanticit...
I found [1], created all the way back in 2021, which seems to first introduce this "kidney disappointment" term, and I would have to assume that perhaps the authors preferred to translate it from their native language.
LLMs did not exist in their current form in 2021, so couldn't have used it to write down a coherent thought, let alone a paper, at the time.
At the same time, I do live in the same country as the authors, and while we don't speak the best English, being a student at a college without some familiarity of the English language to come up with the term "kidney disappointment" is, let's just say, hard to believe for me.
Edit: This other comment[2] made me realize what it might have been: plagiarism avoidance.
Unfortunately, many colleges here do not have the best reputation and often run like degree mills, and the aforementioned comment made me remember a conversation with a friend, who mentioned writing a review paper.
Again, LLMs were not really a thing at the time, so they mentioned how they had to run their paper through a plagiarism detection tool, and then they substituted words with their synonyms. I guess someone really did search for the word "failure" and found "disappointment" as an alternative.
[1] https://www.researchgate.net/profile/Suprodip-Mandal/publica...
In fact, we're increasingly seeing a desire to build the "backdoor" directly into the software, e.g. mandatory age verification, client-side scanning, etc.