XMPPwocky

Why?

This was their third office space, serving as their headquarters before they outgrew it and eventually relocated to Market Square at 1355 Market Street in 2012. The arab spring twitter uprising was fully a CIA/NSA operation.

To be clear, the claim you're making is that because Twitter has their third corporate office in the same building as an AT&T switching center, and US intelligence used a room in AT&T's switching center for surveillance, then Twitter must have been controlled by US intelligence? And thus the Arab Spring uprising, where Twitter was used, was "fully a CIA/NSA operation"?

For what it's worth, this sounds a lot like something downstream from "reward hacking" in ML- in training, passing tests is often sufficient, and thus gets trained for. There are attempts to fix this (e.g. trying to detect such "cheating" and penalize it), but they have their own problems.

Why is something being human or not relevant here?

The difference is that when a human reasoner goes to solve a problem, they'll think "this kind of proof usually goes this way" - following an explicit rule enforcement.

How is this different from "probabilistic pattern selection"?

Hm, say more about what your opinion's based on here?

Hm, it sounds like to you consciousness implies non-determinism, and so determinism implies a lack of consciousness - is that right? If so, why do you think so? And if not, what am I missing?

Yup- the question is "can math be conscious?"

(If you've engaged w/ the literature here, it's quite hard to give a confident "yes". it's also quite hard to give a confident "no"! so then what the heck do we do)

Presumably the decryption key is in the firmware.

I also can't help but note that this blog post itself seems (first to my own intuition and heuristics, but also to both Pangram and GPTZero) to be clearly LLM-generated text.

nit: assuming you mean basis points, one basis point is 0.01%. 4.5bp would be 72.7% to 72.71%. this is 450bp!

How is it a black box? You can get the firmware trivially.

I think your pitch would be significantly stronger if the resources you linked didn't largely revolve around "What if Snoop Dogg had his own currency?"

I'm not ... completely unable to be persuaded that there are use cases for cryptocurrency. But "Ashton Kutcher should have Aston Kutchercoin, and develop some sort of referral-based pyramid-shaped scheme with it for his fans" is... what is that?

What makes you say that? Have you read serious arguments against this view (ideally from academic philosophers, not somebody like me on HN!)?

IGN has accused the game developer for being sexist (based on some poorly translated social media posts) and gave it a low score before properly playing/testing it.

What score did they give it? Because I just checked and they seem to have awarded it an 8 out of 10. I don't follow game reviews, but surely review inflation hasn't gotten so bad that people are on some sort of anti-"woke" bent over an 8 out of 10?

I figured "well maybe they changed it after publication, those sneaky skags", so looked at the ol' Wayback Machine capture for Aug 16th, four days before launch, when the review embargo lifted: https://web.archive.org/web/20240816141422/https://www.ign.c...

8/10 there too- "Despite some frustrating technical issues, Black Myth: Wukong is a great action game with fantastic combat, exciting bosses, tantalizing secrets, and a beautiful world.". Really a horrible thing these journalists have done, ripping into the game like that.

Now, I did also find a report from IGN about 6 months ago that some people claim used "mistranslated quotes". But I don't speak Chinese, and the people online who I've found yelling at IGN about this appear to have used ChatGPT or Google Translate to translate things themselves! So- do you speak Chinese? If not, how do you know the claims of "mistranslation" are correct (again, for a thing that is not a review)?

Could you be more specific? What underclass (and why don't they matter?), and what segregation?

Experts in any field are generally not the ones communicating with, and making recommendations to, the public.

" by the same people that believe whatever they're told by anyone in authority, and then feel the urge to pass it on to others without critical thought."

respectfully, what the hell is this bit doing in your post?

In fact, humans will be running at relativistic speeds within this century, risking the total destruction of the Earth if they should ever trip over something and fall down.

Scary stuff. And it's not science fiction- it's based on real, observed trends and scaling laws. Seems impossible? Well, they said the four-minute mile was impossible, too.

Absolutely agree. When I was a kid, I took a MOOC (MITx 6.002x) because I was interested in EE. They said calculus was a prerequisite, which I didn't know yet, but I went ahead anyways; every time the course required calculus, I'd go read the relevant chapters in Strang's Calculus. Felt incredibly natural, and I ended up going through the rest of Strang just for fun- probably the best learning experience I've ever had, and I doubt I'd have ever done it without having the MOOC there to motivate the problem.

The existence of sentient AGIs would certainly have wide-ranging impacts on the law!

This case is not about sentient AGIs.

But bigger fixed-length instructions mean more I$ pressure, right?

What stops an attacker from uploading their own firmware that dumps out everything in OTP?

I have UI to enter a custom URL for search in Firefox for Android, with no extensions beyond Privacy Badger, Unhook, and uBlock Origin- interesting.

Are the activation functions truly different, or just different parameter values to one underlying function?

I see no sibling comment here even with showdead on, but I could buy that (there's a lot of papers and only so many parameters, after all- but you'd think GPT-4o's search stuff would help, maybe a little better prompting could get it to at least validate its results itself? then again, maybe the search stuff is basically RAG and only happens one at the start of the query, etc etc)

Regardless, yeah- I can definitely believe your point about corpus size. If I was doing, say, frontend dev with a stack that's been around a few years, or Linux kernel hacking as tptacek mentioned, I could plausibly imagine getting some value.

One thing I do do fairly often is binary reverse engineering work- there's definitely things an LLM could probably help with here (for things like decompilation, though, I wonder whether a more graph-based network could perform better than a token-to-token transformer - but you'd have to account for the massive data & pretrain advantage of an existing LLM).

So I've looked at things like Binary Ninja's Sidekick, but haven't found an opportunity to use them yet - confidentiality concerns rule out professional use, and when I reverse engineer stuff for fun ... I like doing it, I like solving the puzzle and slowly comprehending the logic of a mysterious binary! I'm not interested in using Sidekick off the clock for the same reason I like writing music and not just using Suno.

One opportunity that might come up for Sidekick, at least for me, is CTFs- no confidentiality concerns, time pressure and maybe prizes on the line. We'll see.

Every now and then, I'll actually sort of believe an article like this. Then I go and test the current models on things like semantic search.

For instance -

The Hough transform detects patterns with certain structure in images, e.g. circles or lines.

So I'm looking for academic research papers which apply the Hough transform to audio spectra, to recognize the harmonic structure of tonal audio and thus determine the fundamental pitch. (i.e. the Hough space would be a 1D space over fundamental frequency).

I've spent probably 90 minutes, over the several times I've read an optimistic post like this, asking various LLMs (mostly GPT-4o, though my early tests predate GPT-4o, and I've also tried Gemini and Claude), prompts along the lines of

The Hough transform detects patterns with certain structure in images, e.g. circles or lines. I'm looking for academic research papers (please link them or provide a DOI.org link at least) which apply the Hough transform to audio spectra, to identify the harmonic structure of audio and thus determine the fundamental pitch. Make sure to provide only papers that actually exist. If you can't find anything particularly relevant, say so as a disclaimer & just provide the most relevant papers you can.

This is a reliable "fake paper generator", unfortunately - it'll just make up plausible garbage like

Here are some academic papers related to applying the Hough transform to audio spectra for identifying harmonic structures and fundamental pitch:

"An Audio Pitch Detection Algorithm Based on the Hough Transform" Authors: Mark W. and John D. Published In: IEEE Transactions on Audio, Speech, and Language Processing DOI: 10.1109/TASL.2008.2000773 Abstract: This paper proposes an audio pitch detection algorithm that utilizes the Hough transform to analyze the harmonic structure of audio spectra and determine the fundamental pitch.

This paper does not exist. Complete waste of my time. And again, this behavior persists over the >1 year period I've been trying this query.

And it's not just search-like tasks. I've tried asking for code and gotten stuff that's outright dangerous (try asking for code to do safe overflow-checked addition on int64_t in C- you have about an 80% chance of getting code that triggers UB in one way or another). I've asked for floating-point calling conventions on RISC-V for 32-bit vs 64-bit (would have been faster than going through the extension docs), and been told that RV64 has 64 floating-point registers (hey, it's got a 64 in the name!). I've asked if Satya Nadella ever had COVID-19 and been told- after GPT-4o "searched the web"- that he got it in March of 2023.

As far as I can tell, LLMs might conceivably be useful when all of the following conditions are true:

1. You don't really need the output to be good or correct, and 2. You don't have confidentiality concerns (sending data off to a cloud service), and, 3. You don't, yourself, want to learn anything or get hands-on - you want it done for you, and 4. You don't need the output to be in "your voice" (this is mostly for prose writing, for code this doesn't really matter); you're okay with the "LLM dialect" (it's crucial to delve!), and 5. The concerns about environmental impact and the ethics of the training set aren't a blocker for you.

For me, pretty much everything I do professionally fails condition number 1 and 2, and anything I do for fun fails number 3. And so, despite a fair bit of effort on my part trying to make these tools work for me, they just haven't found a place in my toolset- before I even get to 4 or 5. Local LLMs, if you're able to get a beefy enough GPU to run them at usable speed, solve 2 but make 1 even worse...

Thank you, this is very interesting!

Fascinating, thank you!

Do people go on dating sites to look at "hotties"? I've heard there are better websites to do that, many free of charge!

(Not a rhetorical question - as a queer person who's never used a dating site or app and who's been in a long-term relationship (now married) for almost 8 years now, I really do have no idea what people do on there.)