I don't understand why so many of these comments HN is getting are so fixated on writing style. I appreciate that stylistic traits associated with AI-written text are often indicative of contentless slop. But lots of people also write like that. To moan about writing style without even considering the value of the content just sounds cranky to me.
Anyway, I enjoyed reading the experiment, and the starting premise, and the embracing of a fairly mundane outcome. Reminds me of running various generative systems and looking for emergent states.
Yea I'm curious if any of those negative commenters considered that the author is German, and English may not be their primary language — but no, apparently there is a new surefire way to detect AI content. Not forced enthusiasm. Not em dashes. Just too many sentences in a row starting with the word Not.
Because writing is a process suited to some minds more than others. And if your bar to receiving an idea is that it came from a competent writer then you’ll be limited to ideas that reflect a certain way of thinking (and one many mistakenly conflate with intelligence).
Conversation is a different process that may be suited to other types of thinking. And so on.
It’s true that LLMs can enable unoriginal and shallow thinking to be expressed with a false veneer of original thought. But it does not follow that all LLM-written text is the result of laziness nor that it is without value.
I’m just suggesting, come with an open mind rather than judging the whole thing based on a gut reaction to style. That reliance on a gut stylistic reaction is itself lazy and has led many an intellectual community into echo chambers of groupthink.
I don't read Dickens because I can't stand the style despite the rest of its plot and characters. Bad style is a problem to getting into a work. A bad style can make the content hard to read.
PoW rate limiting actually pre-dates PoW cryptocurrency, the former has just had a resurgence recently in response to AI scrapers pissing in everyone's pools.
Post quantum - as in designed to resist quantum computer based attacks under which rsa would quickly crumble. Why do you associate this with snake oil?
It does sound a bit like the famous "military grade encryption" and it's equally (ab)used by snake oil salesmen.
I can't say anything about TutaCrypt's long-term effectiveness except that CRYSTALS-Kyber is touted as being at the forefront of post-quantum cryptography.
I wouldn't call it snake oil, but right now it appears quantum encryption cracking is only theoretical. I'm not sure how anyone can promise to mitigate attacks that haven't yet arrived.
Global Risk Institute... found that the majority of cryptography experts it surveyed believe quantum computers, more broadly, will be able to break anything encrypted with RSA-2048 within 24 hours within the next 30 years.
Most cryptography experts are probably not experts in quantum computers as well.
We already know the algorithm to break RSA with a quantum computer. We just don't have the hardware yet. Nobody knows when the hardware will be available but a lot of entities are working on it.
It's common in cryptography to mitigate attacks that are known but not feasible without further advances in hardware or algorithms. Nobody wants to wait until an attack is successful. That's why NIST is already working on post-quantum cryptography standardization:
Something inspiring about this: "In dedicated Friday afternoon thinking sessions, he returned to the problem again and again over the past decade, to no avail."
Friend of mine worked used to block off his friday afternoons for 'weekly review'. Which was part big thinking, part end of week nap, and mostly avoiding colleagues who had tricky tasks 'needed first thing monday' they had forgotten to bring up before.
Title seems a little clickbaity. It's on-device scanning, which is off by default, and seems primarily designed to help you get content warnings on incoming content.
The UK’s road numbering system feels like it was designed for the convenience of the people in Whitehall filing their maps rather than those using the roads. Giving adjacent roads very similar numbers is the maximally confusing thing to do. For my brain at least - I have to choose between the M73, M74 or M77 and I often get it wrong. It would be easier if they’d reversed all the numbers to put the zone at the end - I feel M37, M47 and M77 would be easier to remember between.
The judge’s point is that if they want to sue anyone for misleading them, it should be whoever sent them the message. He’s not saying they would necessarily win that hypothetical case.
I love how terrible the name is from a branding perspective. It actually makes it feel more legit like a new tech rather than yet another wannabe assistant. And from what I can see, few have heard of Gemini, Bard, Cortana, Copilot, or Jeeves compared to ChatGPT, which has something of the charm of R2D2 or C3PO.
ChatGPT was never intended/expected by OpenAI to be a big deal in of itself. GPT-3 had been a big step up in capability from GPT-2, but of course wasn't getting any attention outside of the ML world since it wasn't something people could actually use. OpenAI built ChatGPT (says Altman) basically just to showcase GPT-3 since nobody else had bothered to do it. I don't think they were expecting the public to be so enthralled with it and actually find so many useful things to do with it.
Is it so terrible then? "GPT" has a nice ring to it, especially considering that it wasn't meant to become a household name - before ChatGPT blew up, only people who were interested in the AI field knew what GPT was at all.
Anyway, I enjoyed reading the experiment, and the starting premise, and the embracing of a fairly mundane outcome. Reminds me of running various generative systems and looking for emergent states.
Shame there's no rss feed to follow along.
reply