HNHacker News
TopNewBestAskShowJobs

unraveller

459 karma · joined November 19, 2019

e9d0c0ef-add5-4b02-a272-33ff142a4fff@anonaddy.me
submissionscomments
unraveller··on OpenAI whistleblower found dead in San Francisco apartment
Web scraping is legal and benefiting from published works is entirely the point, so long as you don't merely redistribute it.

Training on X doesn't run afoul of fair-use because it doesn't redistribute nor does using it simply publish a recitation (as Suchir suggested). Summoning an LLM is closer to the act of editing in a text editor than it is to republishing. His hang up was on how often the original works were being substituted for chatGPT, but like AI sports articles, overlap is to be expected for everything now. Even without web scraping in training it would be impossible to block every user intention to remake an article out of the magic "editor" - that's with no-use of the data not even fair-use.

unraveller··on OpenAI: Sora
Is it better or just more distracting away from it's flaws and using flaws to advantage? I only see repulsive mouth movements to induce fear, face coverings to hide the uncanniness, dreamy physics sim to distract. Not so out of place in present day hollywood but never any coherence of feeling.
unraveller··on Getty Images CEO: Respecting fair use rules won't prevent AI from curing cancer
The claim is around failing fair-use by a substantial portion of the work redistributed for commercial gain at the expense of the rights holder.

The "trained on" isn't the law being broken here, that's the scare campaign. If it wins it's just the smoking gun that helps prove the other points easier, you'll still be able to train on anything.

But if the portion of some work thought to be there is not actually in the final product (prompt engineering a reproduction by detailed description shouldn't count as that is tracing with words) and it's just some style mix (as most AI engineer's say) then I don't think anything will come of this.

unraveller··on The AI reporter that took my old job just got fired
No AAA professionals in the entertainment profession have built a gen AI model up through their own vision that I'm aware of. AAA Nerds build the next version to distract away from flaws of the last and hope the api will get heavy use. I wouldn't expect output to be compelling with that attitude. I expect repulsive or parody clips to continue until some great creatives feel like building their replacements properly.
unraveller··on The AI reporter that took my old job just got fired
People are judging AI by what abusers of AI put out there for lols and what they themselves can wring out of it. They haven't yet seen what a bunch of AAA professionals with nothing to lose can build and align.
unraveller··on How did you do on the AI art Turing test?
Like all ai-or-not tests this fails to keep a similar high quality threshold for both kinds so it intends to waste time not appreciation of either kind of art.

The curator was selecting human output for overlap with ai flaw/artifacts that are likely to confuse at a glance. He wasn't selecting randomly above a high quality threshold for both kinds as implied.

Typically AI is boring, takes the easy way out upon further inspection, likes lone straight lines and face front on shots and it just so happens there are many tests which he found old human examples of this, with large perspective/lighting flaws as well.

I don't care what the point of art is consesused to be, or if elephant-made art is distinguishable from a 5th grader's art.

The turing test was "obsolete" before eliza time, the solution was: it doesn't matter to me because i'm using it as if it were human.

unraveller··on An Uncanny Moat
I don't buy into this "Think of the children, regulate adults now" line of thinking for chatbots.

Good parenting solves these personality trait concerns without a foot in the door for mandatory suspension-of-disbelief busting watermarking in all ai output. Adults have the right to regard any human (or non-human) effort in any sphere at whatever level they want, for whatever reason. They don't need to be reminded of it every interaction forever more as punishment for voice actors losing jobs in 2023 and to lighten the load on parents.

I get that children can't distinguish personhood well and might grow up disrespectful of real humans if left to AI but the same argument applies if left watching youtube all day. There are new expectations on all of us to compete with new tech. We can manage without knee jerk nanny state style encroachments at every turn.

unraveller··on How long til we're all on Ozempic?
Anyone can list 50 no-no things they don't eat and omit the 5 types of foods they do choose that causes the most damage. Volume & variety are not to blame given most sauces are calorie dense. The issue generally is the person eating for pleasure conflicting with their stated aim to eat for weight maintenance.

If it tastes good spit it out. Countless people ready to tell you no sacrifice is needed in that domain but they have something to sell or trade.

unraveller··on NotebookLM's automatically generated podcasts are surprisingly effective
You ever watched a reviewbrah video? he doesn't get to "without any further ado" moment until after the halfway point of the video. The prank is the wasted time. But the joke is every other YTber does it more subversively without you getting any laughs out of it. It proves we give way more attention to slop then we dare to calculate.

https://www.youtube.com/@TheReportOfTheWeek

unraveller··on Australia to Block Internationally Purchased 4G/5G Phones as Part of 3G Shutdown
You'll get the dreaded unsupported announcements before each call if your phone drops down to 3G to make calls. There are ways to force 4G/VoLTE calling for phones (see below) but they can get impossibly cumbersome for even CS Major consumers because telstra wontfix.

Telstra are effectively refusing to allow international phones on their VoLTE network even though most devices have band 28 and are fully capable of emergency calls and the lot.

If your device works fully over 4G but isn't on their approved whitelist they are saying it will be kicked off regardless of 4G capability. Legally they claim to be unsure of your 3G override for calls even though it's assured by telstra enabling their network config properly and one extra step by users of xiaomi phones.

Easy things to try:

  *#*#86583#*#* 
”VoLTE Carrier check was disabled” for xiaomi phones adds voLTE menu item in sim menu

  *#*#4636#*#* 
brings up sim TESTING app (or Tap internal storage 7x) you can pin network to "LTE only" and shows IMS registration status

add a new APN with IMS/IMS/IMS for Name/APN/Type (should happen auto if your carrier has VoLTE enabled for your account so call them to make sure, this is meant to be default as of 2022)

check your phone has the required band 28 for telstra's non-standard VoLTE

Advanced things to try:

Modem Test Tools (Tap kernel version 7x > more menu) > apply different australian .mbn and CARRIER files from xda forums (needs latest OS, root)

latest Qualcomm Drivers plus the latest QPST-Tool windows software to apply .mbn files (requires PC, root then unroot)

otherwise switch carriers for your bands or get a new phone when the ban hammer comes down...

scaring people into overpaying for your phones as they can't tell how poorly they are about to be treated for totally unnecessarily reasons is quite a horrible act.

unraveller··on I Am Tired of AI
First gen models trained on books directly. Latest Phi distilled textbook-like knowledge down from disparate sources to create novel training data. They are all fairly open about this change and some are even allowing upset publishers to confirm that their work wasn't used directly. So stones and ionized water go in the soup.
unraveller··on I Am Tired of AI
If you go back to the earliest months of the audio & visual recording medium it was also called uncanny, soulless and of dubious quality compared to real life. Until it wasn't.

I don't care how many repulsive AI slop video clips get made or promoted for shock value. Today is day 1 and day 2 looks far better with none of the parasocial celebrity hangups we used as short-hand for a quality marker - something else will take that place.

unraveller··on The Intelligence Age
That's a good argument against a full accounting education but not against education itself, it's okay to drop legacy things like calligraphy class from curriculum.

We are never going to automate ALL the work humanity does, that's as meaningless as both parties outsourcing their marriage to someone else. We still need to be well-rounded individuals to audit and direct machines (and ourselves) to make breakthroughs where we see fit.

unraveller··on They stole my voice with AI
I assume Jeff wants cease and desist too as this seems more blatant on the surface. Starts a cat and mouse game until they find a variation they feel is different enough to ignore his pleads. Some will use this new clone tech for free publicity hack and others will claim it's still their voice and try to censor it as punishment for targeting them or not doing a bigger deal for the real voice and finding a better one.
unraveller··on They stole my voice with AI
With the tools I'm aware of you just add clips of as many types of voices you want blended in and it blends everything in them to an unknowable uncontrollable degree plus entropy of the system. I suspect their story is they have added in more pleasant sounding voices to the mix which provides enough differentiation.

Question is: who is to say how much is needed before it escapes likeness theft? The king of generic nerd voices is going to claim excessive likeness and the accused lifter isn't going to reveal his whole process. Also tuning AI voices by ear is surely possible soon so category kings are not saved by demanding to be left out of training. A ministry of voice authority sounds bleak.

unraveller··on They stole my voice with AI
Human voices store shared accent data too not just biometrics, your siblings/twins/sorority sisters likely present with extremely similar voices which can cause the same confusion as AI clones. is it most famous sibling first gets the copyright or are siblings the only ones with the power to supersede it or must they alter their voice? Simple voice never really was guaranteed to be a unique part of identity since it changes so easily over time and ages.
unraveller··on Show HN: Bot or Not? AI voices vs. humans
People speak with a robotic voice when they read. If the stats are to be trusted externally it would need a promise on the page that all human voices have been instructed to perform above the basic level of natural intonation - just like the bots have.
unraveller··on Strange Design Decision: An Oven with French Doors
when placed at head height a downward opening door would impede more for heavy items, even if split in half. Serving short customers is better as both tall & short people can move to either side easily. The split doors just saves room in small spaces and gives options.
unraveller··on LinkedIn is now using everyone's content to train their AI tool
They scraped user data. It's worse than that time political campaigns targeted voters.
unraveller··on Stephen Fry – AI: A Means to an End or a Means to Our End?
Humanity is not a work task that can be obsoleted. It's hard to imagine total human versatility overlap by AI is possible let alone around the corner.

The Globotax for UBIbucks is just Inescapable Socialism by another name, today it solves imagined ills from murky AI predictions, yesterday it solved automation. There is more innovative work to be done and that needs an accurate human feedback loop.

unraveller··on Stephen Fry – AI: A Means to an End or a Means to Our End?
>It doesn’t take much for an Ai to find out that if it is to complete the tasks that are given it, then its first duty (obviously) is to survive.

Or just have another robot fetch the coffee if the first goes offline? Death and anthropomorphism are clearly the wrong concepts here. His error is to imagine households so poor they won't be able to send another robot to see where the last robot broke down. Fry is really out of touch with the way things and parts of things are tinkered with. Even chatgpt4-o1 CoT is not 1 thing but a system sending another agent to see where the last one went wrong and adjust.

So evolution magically explains self-preservation in AI and is sure to emerge in machines like animals and will alchemize more "self-awareness". Fantastic paradoxical sci-fi storytelling but not convincing at all in the real world.

The facts are AI never had any self-awareness, it doesn't know where it begins or ends, neither the system or its creator can tune into actual evolutionary forces (artificial selection might be a better term for pretending to though it frightens less), and it cannot "lie" or "breach itself" with any sense of agency. The only things that will be breached for the foreseeable future are the pundits' fashionably wrong-headed expectations.

unraveller··on Huawei Unveils $2,800 Trifold Mate XT Just Hours After Apple Launches iPhone 16
A trve roman doomscroll device
unraveller··on Huawei Unveils $2,800 Trifold Mate XT Just Hours After Apple Launches iPhone 16
here is actual video of a hands-on unboxing

https://youtu.be/K95X37h44zw?t=37

unraveller··on NaNoWriMo AI Tech Policy
Writer's block is going to be a thing of the past now AI is here anyway so they might as well embrace it.
unraveller··on Procreate defies AI trend, pledges "no generative AI" in its illustration app
The pushback is a rebuttal to the with us or against us moral nature of their pledge. The feature set was clearly contentious enough to publish a manifesto which expects a fracture of userbase. If they've misjudged the numbers I'm sure some minimum of missing creature comforts will creep back in. If only because Art can be more than what the commercial tool makers claim.
unraveller··on Cerebras Inference: AI at Instant Speed
Fantastic demo. Do you know what's the difference between your stack and the livekit demo? [1] it shows your voice as text so you can see when you have to correct it.

Llama3 with ears just dropped (direct voice token input) which should be awesome with cerebras [2]

[1]: https://kitt.livekit.io [2]: https://homebrew.ltd/blog/llama3-just-got-ears

unraveller··on The Monospace Web
I can't stand monospace for reading, so I made a bookmarklet because all reader modes suck leaving font preference even though this site works fine just changing the normal text tags over to sans-serif.

   javascript: (function() {
    (function() {
        const textElements = document.querySelectorAll('p, span, h1, h2, h3, h4, h5, h6, li, td, th');
        textElements.forEach((element) => {
            const textColor = getComputedStyle(element).getPropertyValue('color');
            element.style.fontFamily = 'system-ui, -apple-system, sans-serif';
            element.style.WebkitTextStroke = `0.4px ${textColor}`;
        });
        location.hash = "";
        history.replaceState("", "", location.pathname);
    })();
})();

https://caiorss.github.io/bookmarklet-maker/

unraveller··on Ask HN: Which movies did you watch multiple times?
nu-movies aren't aiming to be loved in isolation, they've got more important corporate boxes to tick.
unraveller··on Overview of the Stanford Creative Writing Lecturer Situation
I would consider the possibility your positions are naturally going to AI by then or effectively sacrificed to those with more DEI "merit".
unraveller··on What Lasts and (Mostly) Doesn't Last
Yes, well done, the canon is winning the most amount of scorn from academics, HOWEVER. The desire for quality will outlive them all, including big tent slop and long-tail slop.
← PreviousPage 3 of 19Next →