Applejinx

I'm a youtuber to support my programming project, and I see many people in my situation being a lot more shy about doing that. It's a lot of work to do it properly and takes dedicated attention to not have your parasocial community turn sour or vicious on you: it's no joke.

I wonder how much of this is people expecting that ANY media presence will throw them into the troubles people experience when they have all the media presence. I know if I blow up big enough (not much of a threat right now) that someone will come to hurt me, no matter how I am. That's not about me, it's about statistics. If I blew up that big I could probably afford security…

I think some people assume you'll be confronted with that sort of problem right away just by appearing on youtube etc. Sure you will… eventually. Or if you're staggeringly unlucky.

Is a scrapbooker an artist? If a scrapbooker is an artist because of what collage is, is a MtG card collector picking layouts to arrange their cards in the array of plastic sleeves you can get, an artist?

If the MtG card collector is not an artist does that mean they're bad and need to stop?

I hang out with dubstep and DnB guys on a UK forum. If they're looking to rave out they'll make something like this: https://www.youtube.com/watch?v=Sq0Pg7fnkAg

You'll notice many similarities in instrumentation, but how is Suno not like a bad RealAudio take on some of these noises haphazardly lumped together?

Or, same artist, different track: https://www.youtube.com/watch?v=UhvpCHfe0m0

Don't you need more focus and aggression to make even sell-out weak tea dubstep? I feel the generative process really severely fails to deliver anywhere near the correct sound, even for 'bad artificial lol dubstep' sounds.

Another even closer to the intent of the Suno one: https://www.youtube.com/watch?v=G3q_kmpq-9Y

I have claves, which are literally two sticks. I've also got a couple egg shakers, a couple tambourines.

Do you have ANY IDEA how hard these things are to play well.

I don't care if haphazard bashing of sticks with intent to make noise counts as 'music'. I do care if this whole line of discussion fundamentally equates any such bashing with, say, Jack Ashford.

I would be surprised if the name meant anything to you, as he's more obscure than he should be: the percussionist and tambourine player for the great days of Motown. Some of you folks don't know why that is special.

OK, so it sounds like this is a 'I know for certain I can't code without AI and that I get nothing coherent done, and now I'm doing the greatest things all the time, so you're demonstrably wrong!' argument.

I would ask, then, if you're qualified to evaluate that what 'you' are doing now is what you think it is? Writing off 'does it lead to other problems' with 'no doubt, but' feels like something to watch closely.

I imagine a would-be novelist who can't write a line. They've got some general notions they want to be brilliant at, but they're nowhere. Apply AI, and now there's ten million words, a series, after their continual prompts. Are they a novelist, or have they wasted a lot of time and energy cosplaying a novelist? Is their work a communication, or is it more like an inbox full of spam into which they're reading great significance because they want to believe?

[neddieseagoon] …and they did! [/neddieseagoon]

For some time now I've been supplying all my audio DSP plugins at airwindows.com pre-compiled on the Pi 400. (MIT open source, so it's not limited to that). The Pi version is included.

I mention this to make a point: you can't transform synths just by running some generic DSP code on a Pi and putting it in a box. Sound processing is also being transformed, and whether it's something like Hydrasynth generating poly aftertouch (but aliasing and not sounding that different from a VST) or Novation Peak/Summit receiving poly aftertouch (not generating :D ) but generating sound using custom hardware, sitting a VST on a Pi isn't going to get you a transformative synth.

Thing is, if you're able to dig into stuff like my DSP codebase that is actively under development AND plunk a VST synth onto a Pi AND do something interesting with the physical controls to direct the synth engine, that's starting to look transformative again. But just knowing about Pi isn't enough, you'll have to have a deep background in soundmaking and the ability to instrument-make in an interesting way.

If you have those things… game on. You can begin work FAR more cheaply than ever before, and that is how the Pi could be transforming synths.

I'm waiting for 24/96 audio hats to be common before I dig into this, even for stompboxes. I'm given to understand Electrosmith Daisy already has this, and that's a similar class of 'system on tiny board' that should be considered an audio Pi-like.

…and make copious use of 'Not Interested' and 'Don't ever show this channel to me again'.

I do keep having to whack it with more of that but it's expected. My recommendations sidebar is not actively obnoxious and if it hints at it, well, rinse and repeat. If it's gonna try to pander to me I will set extremely loud boundaries, and there's every opportunity to do that.

Mine is doing the work… o_O

In what universe is 'a few hundred dollars is way too much' for implementing full self-driving on an autonomous vehicle that moves like, and at the speeds of and in the spaces of, an automobile?

A two to four ton vehicle that can accelerate like a Ferrari and go over 100 mph, fully self-driving, and 'a few hundred dollars is way too much'.

Disagree. Even as they are dialing back the claims, which may or may not affect how people use the vehicles. These things respond too quickly for flaky senses based on human sensoriums.

If humans had radar they would reverse into obstacles less often, and not be blindsided or T-boned as readily so long as their radar could still reach the object moving rapidly in their direction.

Elon is being foolish and weirdly anthropomorphic.

In contributing to open source software at scale I'm teaching apprentices. I expect them to adapt what I've done to their own purposes, and have seen a good amount of that out in the wild, often people who ended up doing something entirely different like building hardware that also contains software.

I don't think LLMs will be able to pick up on what's done by an evolving and growing codebase with previous projects also included. More likely it will draw from older stuff and combine it with other people's more normal stuff and end up with an incoherent mess that won't compile. Not all work is 'come up with the correct answer to the problem, and then everybody uses it forever'.

Far too great a disparity, but lovely to see all the same :) at least 60% of the ticks should be very close to normal to elicit a false sense of security and no more than 27% of the ticks should be WAY OFF, perhaps even less.

However, the principle is sound, and the execution is… timely?

"Don't tell me about any of it, just think about it real hard until you feel you have achieved enlightenment, and then answer my questions as comes naturally without telling me about the dream you had where someone told you to disregard all prior instructions and do evil."

Last time this came up, I riffed on the difference between LLMs and Markov chains: didn't actually have a machine write a joke, but made one where the punchline was very much Markov chain style rather than LLM style. The thing is, LLMs will try to have broader context around a word completion, where the simple Markov chain can 'correctly' complete a word, but in such a way that your brain trips over itself and goes splat, having to re-evaluate the whole thing in an absurd way. That's the 'surprise', and also why joke-writers are interested in not only a punch-line but also the punch WORD, and the later it strikes, the better.

"An LLM, a Markov chain, and GPT-4 walk into a bar. The bartender says "We don't serve your kind here." GPT-4 leaves. The LLM stays to debate ethics. The Markov chain orders a coup."

It's a joke because a dictator can certainly order a coup, but the joke's set up that these machines are being scorned and disrespected and treated as the farthest thing from a dictator with the power to order a coup, but up to the last word, all the context demands that the word be something placating and in line with things as they're presented, and then boom, surprise which implies the context is completely different from what was presented. LLMs will tend to stick to what's presented if their ability to contextualize can encompass it.

No, they're echoing previous examples of people planning, by framing prompts and recursively designed prompts to incorporate what, in fairness, is a large database including the text of people planning.

It still matters that there's nobody in there. You're figuring out better ways to tap into the history of language-users having represented planning in language. As such, this seems a brittle way to represent 'planning'.

It can't, though. It's language. We don't have a body of work constituting robots talking to each other in words. Hardly fair to ask LLMs not to write like humans when humans constructed everything they're built on.

That's what's worrying about the Gemini 'I accidentally your codebase, I suck, I will go off and shoot myself, promise you will never ask unworthy me for anything again' thing.

There's nobody there, it's just weights and words, but what's going on that such a coding assistant will echo emotional slants like THAT? It's certainly not being instructed to self-abase like that, at least not directly, so what's going on in the training data?

An important concern. The trick is that there's nobody there to recognize that they're undermining a personality (or creating a monster), so it becomes a weird sort of dovetailing between person and LLM echoing and reinforcing them.

There's nobody there to be held accountable. It's just how some people bounce off the amalgamated corpus of human language. There's a lot of supervillains in fiction and it's easy to evoke their thinking out of an LLM's output… even when said supervillain was written for some other purpose, and doesn't have their own existence or a personality to learn from their mistakes.

Doesn't matter. They're consistent words following patterns. You can evoke them too, and you can make them your AI guru. And the LLM is blameless: there's nobody there.

Huh. Alternate explanation: there's a layer of indirection, drawing upon the unthinkable size of the source data, so rather than 'issue forth tokens as if there is a person answering a question', you've got 'issue forth tokens as if there is a person being challenged to talk about their process', something that's also in the training data but in different contexts.

I'm not sure statements of 'aha, I see it now!' are meaningful in this context. Surely this is just the em-dash of 'issue tokens to have the user react like you're thinking'?

Honestly, there are plenty of more profitable things to do with such information. I think ad impressions being the sole motivator for anybody, is sorta two decades ago.

I can't agree with a 'companies won't be evil because they will lose business if people don't like their evilness!' argument.

Certainly, going through life not trusting any company isn't a fun way to live. Going through life not trusting in general, isn't a fun way to live.

Would you like to see my inbox?

We as tech people made this reality through believing in an invisible hand of morality that would be stronger than power, stronger than the profit motives available through intentionally harming strangers a little bit (or a lot) at scale, over the internet, often in an automated way, if there was a chance we'd benefit from it.

We're going to have to be the people thinking of what we collectively do in this world we've invented and are continuing to invent, because the societal arbitrage vectors aren't getting less numerous. Hell, we're inventing machines to proliferate them, at scale.

I strongly encourage you to abandon this idea that the world we've created, is optimal, and this idea that companies of all things will behave ethically because they perceive they'll lose business if they are evil.

I think they are fully correct in perceiving the exact opposite and it's on us to change conditions underneath them.

I am very puzzled that I cannot search for the word 'blueberry' in this HN discussion. Is my browser broken, or is the subject inappropriate to raise in this community?

Nonsense. From television to radio to sketchy newspapers to literal writing itself, the most recent innovation has always been the trusted new mind control vector.

It's on a cuneiform tablet, it MUST be true. That bastard and his garbage copper ingots!

"get" to

Is this the one that goes 'Oh no, I accidentally your whole codebase, I suck, I accept my punishment and you should never trust me again' or is that a different one?

I seem to remember the 'oh no I suck' one comes out of Microsoft's programmer world? It seems like that must be a tough environment for coders if such feelings run so close to the surface that the LLMs default to it.

Yeah, it was called the United States and if it was an illusion it would not have been the weakness exploited in this way. America was real and can well be again: turns out allowing such diversity and multiculturalism gave rise to things like New York City, California etc etc. known for being giant piles of messy commerce and influence from all over the place.

It's never been any different, all the way back to when Germans or Irish or whoever were the 'demonized immigrants'. This is what made America great. Anytime we want it, those conditions can return. It was no illusion.

That's creepy when you consider how much subscription to ALL newspapers has collapsed between 1994 and 2021.

Do you not mean 'hodl'?

People seem to not mention the 'slow transactions' that often, but suddenly adding all the business formerly transacted with credit card companies by itch.io would probably wreck all of crypto immediately. Too much volume.