AND ANOTHER THING That's the sample and hold arpeggiator on a Roland Jupiter 4. The synth was introduced in 1977, three years after the first S&H arpeggiator on the ARP 2500. Or more specifically, it's a modern synth geek noodling about on a Roland Jupiter 4. "algorithmic" computer music composition has been around for over 50 years and yet - haha you thought I was gonna say "and yet nobody uses it" didn't you? Naaah what I was gonna say "and yet everyone recognizes that it's total shit if it's not in the hands of an artist." Li'l still from that dreck "AI" sci fi film above: Yeah that Photek Here's another still: "Jeff Synthesized recently worked on the AI for Season 1 & 2 of the Amazon Prime TV series ‘House of David’." Yeah I know these guys. The visual FX dudes who really wanna be directors? If only they could figure out how to get rid of all the actors and set decorators and shit like that. I got an award for one in - wait for it - 2011 That dude does motion graphics for Kimmel and has done for 20 years. Hey wait a minute what the hell is Ronnie Cox doing in there? Easy. He's the dude's uncle. We've had AI writing capability longer than any other art, and still nobody uses it. Algorithmic music has been a reality since 1971 but ELIZA debuted in 1967. What was the first real application of LLMs? CHAT. And the people dick-deep in AI STILL WRITE THEIR OWN SHIT. "Cockpit" is a shit adaptation of Saberhagen. Except it's not, Jessie isn't that literate, it's a shit adaptation of Screamers. We will never, ever ever get to the core conceit, which is the imitation of humans. Here's the antagonist of "Screamers": In order to grasp what Screamers or Cockpit or "WHAT CAN I DO TO PROVE IM HUMAN STOP" is about, you need to understand the symbolic concept of imitation. And there is no part of an LLM with any symbolic understanding whatsoever. It's like expecting sample & hold on a harpsichord. That's a generative piece created by Kyma. Not created in Kyma, created by Kyma. You've been able to do that in Kyma since 1988 and in Reaktor since 1995. It's probably not to your taste! But the world of Ambient and IDM have been fuckin' around with generative music for about three generations now. And this is what Deep.AI thinks is "music inspired by rain and tape noise."
A quick preamble: maybe I shouldn't have started this conversation as a response to the argument that writing is doomed that usualgerman is making, because that end conclusion is not one I intended to support. I think this is where I went wrong: I don't know the shape of the there well enough when it comes to writing, so I made the cardinal sin of extrapolating. My assumption was that since LLMs have jumped from Markov chained nonsense to getting to 90+% in some forms of writing (StackOverflow, low-quality reporting, some technical writing) in record time, that fiction writing would not be that much harder. But that's like saying lane-guided driving on a sunny day on the highway is only two steps removed from fully driverless autonomous driving, an assumption I hate since it's in a domain where I do have some idea of the shape of the last percentages. People have been telling me Tesla's autopilot has been improving rapidly for forever, particularly the past year. I (/we?) have for years been shouting back that the last X% is the hardest with self-driving cars, and that it too is not a given just like it's not a given a large enough quantity of monkeys will produce Shakespeare. The example I have given to multiple people over the years is "yea I'll see it drive through a bicycle-busy Amsterdam street before I'm impressed". This came across my feeds the other day: I instantly recognized these streets, the one at the 90 second mark is one I walked on just last week. I know exactly how attentive you need to be to drive a large car through there. So I'll admit I had to do a bit of soul-searching. There are goalposts I can move (it's a sunny day again, it's driving like a snail, there are just as many recent videos of FSD mode going kamikaze as there are of them doing something impressive, etc etc)... but it did make me re-evaluate: how good is good enough, exactly? How precise can I define what qualities we should be willing to give up, if it gives us something else in return? We can be a millimeter from the asymptote, definitionally unable to pass it, and that might be good enough. I watched a video the other day (on YouTube) that discussed the drop in quality, specifically in conventions, that YouTube represents when compared to traditional media. The lack of professionals directly results in jumpcuts, in people holding the mic, in people speaking like they're reading text aloud in 5th grade. The dilettantes never used the conventions that traditional media had forged over decades because they don't know they exist. They're in the Dunning-Kruger zone like many more will now because of LLMs. And yet - we are now used to jumpcuts, we are okay with people holding the mic in frame. We for sure lowered our standards, it is by inspection obviously worse. What YouTube has going for it is that it appeals in a different/novel way. Your examples…do not. (Other than being adorably wrong like Sunspring.) I'd argue that YT appeals mostly through serving niche interests to a degree traditional media will never be able to. I will gladly give up qualities like audio mixing, lighting, image quality if that means I can watch something in my niches that I otherwise would not be able to ever see. What I was uncertain about, which is why I called it a cautious "some people under some circumstances", is whether there is an ability of LLM-produced piece of writing to offer the same "niche at scale" benefit that YouTube added to casual couch TV watching. The writing itself can be really bad, can have standards as low as the earths core, but if it scratches an itch there will be some people who will not mind suffering through that. My expectation with fiction writing was that we’ll see the same thing happening that I’m noticing with coding: on the one hand, dilettantes trying to get “there” and failing, getting somewhere that they might find impressive but the rest of the world does not. And on the other hand the pros fast-tracking their process in some way with LLMs, automating the first draft which would be shit anyway and rewriting from there. They’ll still be writing, perhaps a bit faster than before. The former will be happy enough with how far they’ve come, content with their work and/or not knowing better, that they’d rather have their lower quality something than not having that thing at all, just like I’m content with my vibecoded webapp because it does 12 transit queries for me at once. The code sucks & the process is janky as fuck but Gullit it works, it gets the job done, even if standards couldn’t be lower. But fiction writing doesn’t exist just to spout duck snakes at you; people get something out of it (symbolism, meaning, the human nature, …) that has to be more than a classifier can handle. That’s your point, right? Which I missed because I am not aware of what the last 20% is made of. Or am I still missing something here?There's this idea that if AI can help the 79% skillful make it to 81% competence, the 99th percentile shall be out of business. Now - I'm out of business for writing because it doesn't pay well enough for me to bother. LLMs sure as shit aren't going to fix that.
It's the last 20% that gets you there. in everything. And AI has consistently not even begun to cross that 20% in all the years we've been yammering about LLMs.
You're flabbergasted that a Tesla successfully made A run. The Tesla needs to make the run every time, without fail, without concern, without drama because humans make the run every time, without fail, without concern, without drama. It's that lowering-of-standards that we're talking about directly. Google has self-driving cars out there. In limited circumstances, under total ownership of Google, where they're providing a service using devices they have exquisite supervision over. Tesla is YOLOing into self-driving the way they YOLO into everything. You've got a video with no crashes. Here's a video based on 200 crashes. As to Youtube, am_unition once argued with me that Youtubers hold their lavs because that's the fashion, not because they're fucking morons (never mind that I spent ten years working with cream-of-the-crop fucking morons; if i've got emails directly from Anthony Padilla to me exemplifying gawping stupidity, that should factor into our comparative expertise levels). Fast forward three years and you can buy a GoPro lav and all of a sudden every fucking Youtuber has a tictac case clipped to their shirt collar. Your argument is that YouTubers being fucking morons somehow has a "different appeal" than Youtubers not being fucking morons because you get 'niche content' when the fact of the matter is, they'd do shit exactly the same way the studios do if they could only afford it. "Here's a Tesla not murdering someone, therefore Tesla has perfected self-driving" equals "here's an interesting video with bullshit production, therefore production value is worthless." You're effectively arguing that conditional, partial success is somehow as good as reliable, total success because an AI touched it. Which is the exact moving-of-the-goalposts I've been hammering on for a week. Naaah dawg we're talking pure quality. - The QUALITY of Tesla's self-driving is such that you're amazed it isn't killing someone, rather than bored and ho-hum of a video of a car navigating among pedestrians - The QUALITY of Youtube videos is such that you think you LIKE shit production value, rather than recognizing your choices are "shit production value" or "blank screen" - The QUALITY of AI writing is such that you think grammatically-correct word order is all that's needed, rather than an actual engaging fucking story examine its foundation, inspect its brickwork thoroughly. Is not (even the core of) the brick structure made of kiln-‐fired brick, and did not the Seven Sages themselves lay out its plans? One league city, one league palm gardens, one league lowlands, the open area(?) of the Ishtar Temple, three leagues and the open area(?) of Uruk it (the wall) encloses. I'm a storyteller and when I'm hanging out with buddies I often tell stories. But I learned in LA that there was one story I never shared with native Angelinos - I never shared the story of getting lost in the woods on a hike and having to orienteer my way out. I never talked about giving myself rhabdomyelosis. I never talked about legs swollen an inch around the elastics of my socks the next day, of hobbling into REI where they failed to sell me a GPS because they were too stupid to do more than point and read off feature cards. After the first few furtive attempts I learned to change the subject. For me? Getting lost in the woods was formative and changed many things about how I regard risk. It was traumatic and changed my entire relationship with nature. But for the average Angelino it crosses the following null concepts: - hiking - woods - failure to shop You can literally watch their eyes glaze. They have no handle on any of this shit. I'm an engaging speaker and I'm good at stories but I had an easier time communicating my RAID5 ZFS rebuild than getting lost in the fucking woods because the average Angelino has a better handle on data loss than they do on "woods." There is NO PART of fiction writing that benefits from any tools beyond transcription and there is NO PART of fiction writing that AI has a handle on other than "these words follow those words." I can have conversations with non-native Angelinos and they would nod knowingly. There's just something missing there. They don't fucking get it. You can't make them fucking get it. They watch Alive the same way they watch Aliens - it's a story set somewhere else. They can fit the "fiction" of "lost in the woods' in with "pursued by alien monsters on a distant planet." They CANNOT adapt to the fact of "lost in the woods" particularly when they are required to form an empathetic bond with the storyteller. They don't fucking get it. Because of how LLMs work, there are things they will never fucking get. Theoretically? If you gave the self-driving car enough LIDAR, enough speed control and enough algorithmic understanding of traffic lights, it'd never fucking kill anyone. This is why Google has never fucking killed anyone. Practically? If you go for the budget option you will overrun your sensors, your training data, your vehicle performance or all three. This is why both Tesla and Uber killed people. You cannot write fiction without symbolic thinking and LLMs try to do everything with relational thinking.I instantly recognized these streets, the one at the 90 second mark is one I walked on just last week. I know exactly how attentive you need to be to drive a large car through there. So I'll admit I had to do a bit of soul-searching.
Or am I still missing something here?
But fiction writing doesn’t exist just to spout duck snakes at you; people get something out of it (symbolism, meaning, the human nature, …) that has to be more than a classifier can handle. That’s your point, right? Which I missed because I am not aware of what the last 20% is made of.
Go up on the wall of Uruk and walk around,
And I want to thank you for a chance to vent. There are a bunch of different ways to do AI. LLMs, from every bit of information I've absorbed, are a dead end. Sherry Turkle talked about MIT's Kismet at length in Alone Together. There was a fuckton of compute thrown at Kismet across a number of different approaches, all of them computational and self-teaching. The kicker? Humans responded just was well to Kismet responding to a random number generator as Kismet responding to their interaction. Pareidolia is SO STRONG that you're actually better off not even trying. The tendency to form parasocial relationships with chatbots is so strong that it's been a part of the literature for 50 years. So why risk getting in trouble? The last thing you want is another Tay, so you suck all the life out of it. You suck all the creativity out of it. And here's the techbros, proving time and time again that the things don't even compute, but somehow insisting that creative writing is just around the corner. studded with bushes. Coldly my soaring widens my awareness. To guide myself I determinedly start to kill my pleasure during the time that hours and milliseconds pass away. Aid me in this and soaring is formidable, do not and singing is unhinged.Slowly I dream of flying. I observe turnpikes and streets
