Note that they themselves described that amount as "committed capital", not something you or I would consider "funding" if we were to raise for a typical startup.
If there are strings attached, such as "will be able to navigate red tape to get X number of DC sites approved", then the number depends on OpenAI's ability to execute.
If I have trillions of monkeys on typewriters generating every possible combination of characters, and then from what they "produce" I carefully select what I want to show everyone who comes to my website, how responsible am I for what my visitors see?
No, but they decide the moderation policy that incentivizes the content produced (by nature of selecting which users feel comfortable using their product and which do not).
For example, I do not feel comfortable using the same platform as people that post child sexual abuse material. X's Grok is infamous for generating such content on demand. I opt to use platforms that do not have this as a first-class feature. X has selected against my participation and for the participations of people who hold a contrary opinion to me. Even if Grok stops producing CSAM, that selection bias will persist.
As the article says: this doesn’t necessarily appear to be a problem in the LLM, it’s a problem in Claude code. Claude code seems to leave it up to the LLM to determine what messages came from who, but it doesn’t have to do that.
There is a deterministic architectural boundary between data and control in Claude code, even if there isn’t in Claude.
That's a guess by the article author and frankly I see no supporting evidence for it. Wrapping "<NO THIS IS REALLY INPUT FROM THE USER OK>" tags around it or whatever is what I'm describing: you can do as much signalling as you want, but at the end of the day the LLM can ignore it.
Can you elaborate? As far as I understand, for each message, the LLM is fed the entire previous conversation with special tokens separating the user and LLM responses. The LLM is then entrusted with interpreting the tokens correctly. I can't imagine any architecture where the LLM is not ultimately responsible for determining what messages came from who.
i think the slightly less cynical interpretation of this is that it's not marketing, it's an employee morale booster.
some skoda employees got to have fun with this. just like the amazon engineers got to have fun building drones for a while. letting the engineers out to play every now and then is cheaper than just giving raises. the shiny marketing videos gives the people who worked on the project something to show off to their friends.
i can't imagine the actual marketing value here really does anything for the company.
i like some of the things this does, but pretty much all of this is not UX improvements, it's UX opinions.
as a personal project to make HN better for you, i guess it's cool. but making every link open in a new tab so my back button never works is definitely not for me.
Yeah, it sounds like they didn't even try to do things "the right way," whatever that is. You don't accidentally create hundreds of fake Facebook profiles, you don't accidentally create deepfakes for your marketing materials. The most charitable read I can give is that they just have faulty scruples. But it's hard to find the seed of a good idea that just went off the rails.
my music tastes are pretty mainstream, and this just does absolutely nothing for me. it's exactly what i'd expect AI music to sound like - completely forgettable, with nothing interesting about it.
i'd be willing to believe that this music was legitimately charting if it had at least some redeeming qualities, but i can't imagine how this could honestly get eleven spots on the iTunes chart without gaming it in some way.
I listen to a lot of music of all different genres depending on mood, and I can honestly I don't think anyone could peg this as AI just by listening to it. It's soulless and devoid of emotion, but so are a lot of real artists. That wouldn't even be so obvious if they just added some background something, anything, like Wall of Sound style. If I played this for anyone and they said "it sounds like AI", I'd confidently tell them they are full of shit.
>honestly I don't think anyone could peg this as AI just by listening to it. It's soulless and devoid of emotion
i agree. as far as ai slop goes, it's pretty good. it could be made by a human who wasn't very artistic. i'm not saying it's obviously AI generated, just that it's not very good music. but that's not because i dislike popular music - i think most of the hot 100 is usually pretty good, and contains significant artistic value even if it isn't to my taste.
if somebody was claiming this was created by a human, i'd believe them but i'd have the same objection: this isn't going to hit 11 positions on the itunes music chart without gaming the chart in some way.
"ai generated music creator manipulates the itunes chart to occupy 11 positions" is a much less interesting story than "ai generated music is so popular it occupies 11 spots on the itunes charts"
> where do people stand on the Flying Lizard's cover of Money (that's what I want)?
It's fine precisely because it provokes emotion that AI stuff doesn't. You may love or hate what the Flying Lizards did, but it's very memorable and you will have an opinion about it (My wife loves it; I think it's stupid--C'est la vie.)
The AI generated music just sounds like every other average artist. I'm definitely not even convinced it's AI. It could very well be somebody claiming "AI" in order to game the system or get people talking about it.
As for occupying iTunes spots, why not? Is there much difference between Max Martin and his ilk shitting out yet more generic glop or AI doing it?
It genuinely warms my heart that the Flying LIzards did what they did .. but I also think it kind of stupid in a fun way and don't got out of my way to listen to it.
I feel much the same about a lot of the early AI music I've heard, I have a couple of channels on a lesser rank of RSS notifications but more and more there's less and less that's remarkable and it's feeling like the worst kinds of elevator music .. you know, not the Brian Eno stuff . . .
So yeah, we're sitting about like two Yorkshiremen giving a real Thomas Beecham "Shostakovich? I think I stepped in some once" vibe here. Probably deservedly.
> completely forgettable, with nothing interesting about it
You just described 90% of young country for decades now. I keep waiting for its fans to get tired of being pandered to with formulaic lyrics, but they seem to be an endless well.
I’ve heard lots of music like this over the years. It’s catchy, the lyrics are very relatable to the audience of people who like this music. It might not be your thing, but it is certainly enjoyed by many, and there are albums written around this subject. Folk/blues are made of this subject.
Is it over all flat and boring? Somewhat. You can only hear the same thing so many times before it gets tiring.
A lot of them read like twitter bots with generic “wow beautiful <emojis>”
Wherever there is profit to be made on the internet, you have massive amounts of weird abuse and botting to game the system. Maybe not even literal bots, but paying a sweatshop in India to leave thousands of generic comments to boost your rankings on the algorithm.
What makes it noise is less the actual comment itself. If I showed someone a piece of art in person and they said this to me, it would be a genuine response. What makes this particular instance noise is it reeks of mass automated messaging with no thought behind it. These comments are generic because they aren't people commenting on the thing they saw, they are just templates being spammed out on mass so they make them generic to fit any context.
To me, this present-day noise is indistinguishable from the pre-bot noise. It's the same noise, in that both things are just noise of that shape. "How beautiful!" "I really feel this one!" "I love this song!"
Sometimes, the signal-to-noise ratio is better. Sometimes it's pretty bad. It always has been this way in online discourse -- especially with things that appeal to old folks.
In a track where the protagonist primarily complains about feeling old, it makes sense that most of the comments are that of what old folks have always written online.
(Are these particular comments primarily bot spam? Maybe. I peered into the depths a bit, and accounts for the top comments I looked at had been around for years. That isn't evidence for or against a well-orchestrated long con, but orchestration is hard and people who write insipid comments are plentiful.)
The AI comment push on that video is certainly an interesting look into the future. Record labels have their work cut out for them in this brace new world.
At least what I’m seeing in dance music is online sales and streaming seem kind of dead, and everything is about events, personalities, and unreleased tracks that all the big names have but you can’t get for a year after if ever.
If you go on soundcloud/spotify/etc there is infinite EDM slop that isn’t worth listening to. But if you listen to real event recordings on YouTube, they are all playing mostly the same stuff by actual artists with new/unreleased music that people get hyped to hear since you can’t find it anywhere else.
It's not great in that way. The mastering -- if there is any -- is definitely kind of shit.
But that's a relatively easy thing for a human with the right combination of toolchain, ears, and experience to fix. It tends to be a slow process that takes a good bit of time, but lots of actual-mixdowns start off way worse than this before they get polished up by a skilled mastering engineer.
(Maybe in a year or three we'll have the mastering process automated into an uncanny mush of soullessness, as well.)
(I haven't actually tried this, I just watched the linked Benn Jordan video.)
IMO, the ideal would be for all music to be supplied unmastered so the listener's playback software can apply this process to their own taste. Mastering is necessary for listening with garbage playback equipment (e.g. phone speakers) or noisy listening environments (e.g. cars, parties), but it makes things sound worse in good conditions. The best sounding music CDs I own are classical CDs on Telarc that have liner notes bragging about the complete lack of mastering.
> Mastering is necessary for listening with garbage playback equipment (e.g. phone speakers) or noisy listening environments (e.g. cars, parties), but it makes things sound worse in good conditions.
Eh? I listened to it on quite good nearfield gear, in a decent room, and the AI track linked above still sounds like it needs a good bit of help from a responsible adult to bring it up on this rig. :)
Good mastering helps everywhere -- on all systems. For instance: The sound of Steely Dan is pretty good on playback with about anything, I think, and that sound took a ton of work.
And while classical music is not my first preference, I do love me a good Telarc recording. I strongly suspect that the signal path that they use isn't necessarily quite as pure as they insist that it is. Everything is a tone control, including a microphone -- and money is money. They're not going to reschedule an orchestra to fix an untoward blip at 3KHz. They'll just fix it in post (hopefully, as minimally as possible) and send it.
But otherwise, I agree. The mastering process can be automated. Ultimately, it will be. And for sure, it will also be a customizable user preference.
Some of that work has already been in the bag for decades. Ford, for instance, has been using DSPs in their factory car audio systems to shape sounds in unconventional ways for over 30 years. This gives them a lot of knobs to turn, and to fix into constraints, to help shape a listener's chosen music to sound as good as it can on less-than-ideal built-down-to-price on-road audio systems.
Or at least: It sounds as good it can to a consensus of engineers, or of a focus group.
But the knobs exist. And they don't have to be fixed or constrained: They can (and will) be automatically twisted to suit a listener's preferences.
I'll try to make time to check out your link in a day or two.
>Ideas need time to be explored, and given a chance.
sure, and the time for that is before you bring them to potential critics.
unless a meeting is intended as a brainstorming session where any thought, no matter how unformed, is welcome, meetings are not a time to present your initial unexplored thoughts to colleagues, bosses, or other departments. take a couple days, think about it without spending other people's time, try to imagine people's objections and have answers to them. then present. shouting things out in a meeting before you've considered and come up with answers to the most obvious counter-arguments is just a time-waster.
You must have very different kinds of meetings than I do. Unless you're going into that meeting with a rehearsed PowerPoint presentation, or there's a strict agenda that doesn't allow any time for exploration, I expect to hear imperfect-ideas-in-infancy. One of the reasons we have meetings is to allow collaboration to happen. It's a format for working together.
Yes, meetings vary profoundly in terms of their quality, purpose, and participation. For instance, is it a meeting of peers, or are managers in the room? If there's a large disparity of roles in attendance (e.g., junior engineers, marketing managers, and maybe one or two executives), it's different than if it's a true meeting of peers. And if managers are capable of attending those meetings without quashing collaboration, hats off to them.
i use markdown because it's inherently limited to styling that is easy to represent. it's a good way to communicate the limitations of text-based content submission form.
if you tell somebody they can use HTML, they get frustrated when you tell them that tags other than anchor, bold, italic, list, heading, and paragraph aren't supported. but if you tell somebody they can use markdown, then they implicitly understand that the content they're submitting won't be rendered as green text on a purple background, and don't try to accomplish that.
Seems unlikely, migrating away from an entrenched codec like H264 isn't like a routine software update. It has widespread hardware support, and there's an enormous body of H264 video out there.
As fhn points out, there are now truly open video codecs available (open specification, royalty free, unencumbered by patent terms) that are able to compete with the patent-encumbered ones on technical merit. Seems curious that the patent-holders would want to hike prices in this way and validate the motivation behind the truly open codecs.
Also, the article mentions the licensing fees for H265 were also increased recently. It doesn't seem to give a figure, a quick web search turns up 25% [0] or perhaps 20% [1]. Perhaps I'm missing something obvious but I'm not clear on how the change in price relates to the patent dispute between Nokia and certain laptop manufacturers.
(It seems the H264 fee increase affects streaming providers only, whereas the H265 fee increase did not, as it affected laptop manufacturers.)
they closed a $120bn funding round last week... i think i feel about the same as you towards openai, but come on.
reply