Probably LRO vs. Danuri

Per the statistics, police shoot and kill quite a few white men as well. Nobody gets worked up about those dudes.

OTOH, when you dig into the individual incidents, pretty much all of those dead white men were actively attacking the police, usually with a weapon (), so it's difficult to get to worked up about those deaths.

Right, this is an indictment of policing in America, not Flock.

But since you clearly cannot trust the police, you double cannot trust them with access to Flock.

You cannot keep the police from getting access to Flock, therefore the only reasonable thing to do to Flock is destroy it.

Yes, this is only* because the police are corrupt, but since they are...

* Yeah, and some other people too, this hyperbole is for the purpose of the discussion in scope

I think we both know that ship has sailed

We don't know the exact answer yet, but IMO, you're two-halves right.

The key difference is, humans aren't "trained" primarily via a clean library of tagged data (which in itself is a learned skill). We experience the world as a more-or-less continual stream of noisy data from our several senses. We self-train how to interpret (and primarily, filter) sensory data; and more importantly, how to derive new training data from it.

To be clear, I'm not saying we have so many neurons to support input filtering (though that's admittedly at least partially true). I'm saying we virtually can't run out of new training data. Every second we're alive, it just keeps coming in.

Good thing we hardly are made out of any bacteria.

Wait...

In which subfield is development not happening rapidly?

Won't you want those things to be software upgradable?

Except instead of pursuing policies that would help open models at the expense of Big AI, anti-AI people pursue just the opposite - policies that can cripple open source AI but have no realistic ability to meaningfully hurt big AI. It's so short sighted.

Actually, no, what they usually do is much more insidious: fingerprinting rather than watermarking. The fingerprint isn't actually included in the audio, it's included in a database. If they want to tell if the track was generated, they just try to match the fingerprint in the database. Can't filter it out of the track like you can with a watermark.

The two mainstream options (both about half a year old now, we're due for something better) are Ace Step-1.5 and Stable Audio 3 Medium.

Ace Step has full vocals with its music. See here for examples. The downside in my opinion (judge for yourself) is that it has Suno's flaw of sounding too "clean", "mainstream" and "uncreative", except even moreso (Udio was always much better than Suno at this, albeit "less well behaved" - but Udio is out of the game now).

Stable Audio 3 doesn't do vocals, though it has other useful features, like inpainting (great for fixing glitches in recorded tracks for example). In my view, it sounds a lot better. It can also be used to create sound effects. License is a bit more restrictive, though, if you actually care about that, but generally won't affect the average user.

Both are trained on fully licensed data and both are open-weights. So even if the devs released a version that included some sort of watermarking in the future (or someone developed a fingerprinting algo for the existing versions), you could just finetune them to break that.

TL/DR, if you want open source and want to just churn out a full track: choose Ace Step 1.5.
If you want open source and want to do instrumental tracks or to supplement manual work (such as your own vocals) or do audio editing: chose Stable Audio 3.

Current top models are already undertrained by multiple orders of magnitude vs their number of parameters. Not for lack of will or money, but for lack of data - We're already training top models with effectively the entire digitally-available corpus of human knowledge and need about a thousand times more.

That said, the huge Chinese models we've seen lately aren't single models, they're MOEs - Effectively a hive-mind of many smaller models working in concert. Bytedance hasn't mentioned that detail yet but it would somewhat mitigate the training issue. This is still absurdly over-spec'd though. There's simply no way to train it any more than all the rest starving for data.

Not just this month - They also had to revise May and June down by 103k.

It's almost like firing all the people responsible for compiling exactly this data and replacing them with political shills (if at all) was a bad idea. Whodathunkit?

No they didn't, because Congress set the limit by statute and did not give the FCC authority to override it. This isn't complicated but you're still getting it wrong.

Some give him planes, some want him to build a resort on top of their genocide...