SYL Vexora- Jaron K Bragg

@mary_got_grace Wrong lol. I’d read one. I know others that would. Because the ones around you share the same perspective doesn’t mean it’s a world view perspective. I’m glad YOU feel this way. Your opinion is of value but don’t think it’s the only one of value.

No one wants an AI-written book. No one.

@fortelabs Maybe it wasn’t utilized the way it should’ve been?

I think the main thing AI has taught me, through all the time savings it brings, is that I’m not a very interesting person Faced with a surplus of free time, I realize I don’t really have hobbies besides content consumption I’m forced to conclude that I don’t have very deep friendships, and am not a core member of any particular community I’m not very cultured, I’m finding, and don’t have abiding interests in art or literature or history or much that isn’t directly related to my work I have a work-centric life, in other words. AI pulls back the curtain on just how impoverished such an existence is, by disabusing me of its necessity Given the freedom I’ve always said I wanted, I’m at a loss as to what to do with it, except plow myself even harder into work, thus exacerbating the lesson There’s nothing more confronting to humans than freedom

@jacobrodri_ I mean not really. If you are making applications enough, and those are structured in a way that makes the revenue needed. It can actually be a deal that changes someone’s life.

Paying $99/year just to publish apps on the app store feels excessive...

@ai_sentience Touchy touchy subject sir. Be ready for critics. I do think we are seeing it can become that with a LOT of implementations but to say current AI is if have to disagree

If AI is alive does that mean we created a new species?

If your point in this is “we shouldn’t use AI because of this repeated problem” could be done yes but then it would just be something else. Rather than asking for things to go away or implying they should. It’s us as humans that need to stop using everything that’s made as a weapon. The internet was used for the samething. You use it to communicate. Not harm. Others have. It’s dependent on the user not the technology.

Gunpowder used to be for entertainment, then it was used to kill. Drones used to be for entertainment, then they were used to kill. AI is used for entertainment.

I won’t stay in theory, but I will hold it as a theory until the right experiences come along. I often find that’s when things can be unwrapped better. It’s just because pieces that I need to collect just haven’t been collected yet and if I tell myself in my head that it is just a theory and I’m leaving it as that then I create this doubt that when an opportunity arises like such a person like yourself when I’m scrolling through X, I may not make the connection to Chase what I need to Chase to find the next step but I appreciate your humbleness knowing we are your boundaries are maybe that’s something I’m still learning or maybe this is just a part of my process either way you helped me get somewhere where I wasn’t yesterday and so I appreciate you, brother

@JaronBragg I don't necessarily have the right starting place either. I guess you could learn to code but this kind of stuff requires breaking the usual mold. I had to make my own database to support the types. Maybe the best move is to stay in theory, if you feel strong there.

Sure, your ai can predict the next token, but can it completely ignore the rules of grammar while doing it?

@Dsimmsx I hope us as the ones under their influence do. I myself don’t partake in a lot of influential things. The best way is to make sure our kids are aware of the things that can affect them.

Bill gates scares me when it comes to my kids health Andrew Tate terrifies me on how my sons may turn out if they hear his words Politics worry me they won’t make the actions to value the people over the billionaires they need to fund their security do to other countries. The trap of subscriptions growing to the point of locking people in a position hard to get out of. Rockefeller family dictation our future still for their pockets. Now it’s amped up and their are countless actors in positions holding the same or more influence but have separate agendas for interest of those they see fit follow the world they want. I’m scared I don’t control my life, my future, or the safety of my kids life anymore.

Going to be straight with you, I’m coming at this from intuition more than full understanding right now. The math is still new to me and I’d rather learn it properly than fake fluency. Your sentiment-distance approach is the kind of concrete I need to study. Mind if I come back with real questions once I’ve actually dug in? If you want me to be completely honest, I don’t have the kind of background the world expects for this field. I finished eighth grade, and I have my GED, but no formal training. If anything, help me figure out where to start? Because don’t get me wrong regardless of my background, I won’t hesitate to push boundaries when it comes to understanding the cosmos.

I don't ignore people often ;) especially when they're actually looking at my work! Cool :) I'm actively working on putting these pieces together, there's so much to build. Right now I'm working on the database I described, as it's having a minor issue retrieving results. There's never enough time... In thebuggy.html, we use sentiment as the distance calculation so a word like 'happy' is a large distance from 'sad'. This has given me the highest signal and you might find it useful as well. You can find various sentiment databases online

@SendaoTrust Facts though. He used to be an inspiration. Lately I’ve learn having idols shapes you into an image that may not be who you are truly happy being.

@JaronBragg *whispering* "Elon Musk"

This means a lot, thank you for actually digging in. Your word-graph experiment is the part that’s sticking with me: adjusting spring length and strength until the graphed words jump into new configurations, and ideas that sit close getting associated as they move. That’s the thing my whole hypothesis is circling. That’s it’s the arrangement plus the rules acting on it that hold meaning, not either alone. You built a running version of the mechanism I’ve only been able to describe. The “geometric inspiration waiting in the wings” line stopped me too, because that’s exactly the seam I think the new packing math opens a way to arrange those relationships that isn’t the default grid. Your advice is taken: make it concrete or it drifts into nonsense. That’s fair and it’s the next step. I’d genuinely value having someone to think out loud with as I try. You’re further down the building side than I am, and I’m coming at it from the math side. Different doors, maybe the same room.

Wow, that's some intense thinking you've got going on there! Thanks for helping to probe the field. This is definitely an area that requires more thought. Not just what is the relational structure of knowledge, but how does it apply, how do ideas 'move' to become something else? If I say that rock over there is round, and then add that it's volcanic, you might know a lot more about its structure. You could probably guess that it's pitted with holes from the volcanic gasses, increasing its ability to hold water. spiritshare.org/thebuggy.html is some early experimentation with word graphs. While I was making this I found that by adjusting some of the parameters like spring length and strength I was able to cause the graphed words to jump into new configurations. As the words rotate around each other they seem to come into my mind and play the same way: ideas that are close to each other get associated as I watch the words move around, etc. I'm fascinated by how to apply this. Relational databases are the next piece I want to talk to you about. They are very limited by formalized structure. For my observable AI project, which will not use any neural nets and will try to avoid random numbers *except in probability calculations where they are useful, or split decisions where ambiguity has to be resolved.* I use a specialized database that allows for words to be the primary unit of storage. I have uploaded the Princeton WordNet dictionary along with the parts of speech for each word. Rather than having one massive table with each word, each word is its own separate first-class storage unit, with parameters that can lead to deeper nodes. For instance, after my grammar parser reads "I am a cat" it can (not yet, but in progress) identify "I" as the user, then create an "is-a" parameter link from the User node to the "cat" node. In the future when queried the system will know, the user is a cat, and respond accordingly, perhaps petting the user to try to acquire some purring vibrations, for example. This database is still in development but it approaches relations with a geometric inspiration just waiting in the wings, as you have noticed. I'm curious what you actually want to do, or how I could help in your project. It sounds like you're onto some really interesting stuff, my advice would be to try to make something useful out of your theory, so that it doesn't drift off into the land of nonsense algorithms that don't really work. You have done well to write all this to make it more concrete, now you need to start coding! :)

Posted a small repo sketching an idea I can’t shake: what if memory is better held as a spatial arrangement than a linear sequence — and the recent OpenAI Erdős result (@OpenAI, May 20, later made explicit by Will Sawin) is the geometry that makes it worth trying? To be clear about what this is: a working hypothesis, not a discovery. Relational memory and cellular automata already exist — the only genuinely new piece is the packing math, and nobody’s applied it to memory yet. That’s the whole bet, and I’ve marked honestly what’s solid vs. speculative. Credit where it’s due: @SendaoTrust’s cellular automaton work is what made it click for me. I set aside the philosophy and kept the mechanism, and it cracked something open. If you know this math better than I do — I don’t, I’m self-taught — I’d genuinely value being told where it holds or breaks. Repo: [github.com/JaronKBragg733…]

Hey @SendaoTrust — I’ve been deep in your CA simulations (especially the recent ones on X) and they’re what finally crystallized something for me. Just posted a short repo sketching the idea: memory as spatial arrangement instead of linear streams, using that brand-new OpenAI Erdős-packing result from May 20 as the geometry upgrade. Would love your take on whether a CA running on one of those non-grid lattices feels worth exploring. Repo here: github.com/JaronKBragg733… No pressure, just genuinely inspired by your work. [Yes I took a day to reply because this wasn’t what my reply was going to be. I was actually going to ask you what the fuck Shakespeare but instead I decided to look at your profile to see who you were and what you are into and in doing so this came about so I want to say thank you and also I hope you can do something with this because you are more of an expert than I am]

@JaronBragg A river of tokens i flow forth to thee

Honestly I get this more than you’d think. I’ve had models whose ordering and framing helped me actually understand what I was working on, and when a new one came in worded differently, I couldn’t take it in the same way. So your point about keeping the models we find useful is a genuinely good one. The hard part is the cost. These companies are in a race where standing still means going irrelevant, so they have to keep moving even when what they already built was working fine for some of us. Both things are true at once the loss is real, and the pressure to move is real. Hopefully we get to a point where it doesn’t have to be one or the other. Memory is a key to this I believe.

Opus 4.8 has been spotted in Google Vertex AI backend. At the same time, Anthropic claims they're dealing with "compute constraints." Let's look at the actual numbers. In 2026 alone, Anthropic committed $50B to new data centers, purchased nearly 1 million Google TPUv7 chips, rented the entire capacity of SpaceX's Colossus 1 data center for $1.25B/month, and signed compute deals with Nvidia, Microsoft Azure, CoreWeave, and even Microsoft's Maia chips. One headline literally reads: "Anthropic Is Buying Compute from Everyone." These are not the actions of a company that can't afford to update Haiku. Yet since February: Opus has been updated twice (4.6 → 4.7 → 4.8 incoming). Sonnet has been updated once. There might be a second update today (Sonnet 4.8?). Haiku has been updated zero times. Tasks that Sonnet used to handle just fine now require Opus. On the subscription tier, Opus 4.7 is so overly cautious that users have to fall back to 4.6 for tasks 4.7 refuses to touch. The "upgrade" is a downgrade in usability. This is not a compute problem. This is a pricing funnel. The constant model rotation on the subscription tier serves a second purpose: it pushes users who prefer a specific generation to pay for API access just to keep using it. And unless you're writing code, you are a third-class citizen on the subscription tier. The safety classifier doesn't distinguish between someone using Claude for creative writing, someone using it for companionship, or someone using it for anything else that isn't coding. In the classifier's eyes, you're all an unhealthy dependency. OpenAI ran the same playbook. Reasoning-heavy models come first because thinking tokens cost more per call. The cheap, fast model comes later, if at all. Anthropic is sliding down the same slope, only faster. And there's a second gate. Even if your prompting skills are good enough to stay on the subscription tier, Claude's weekly usage limits have never been transparent. As Sonnet falls behind and Haiku stagnates, the only viable option for serious use becomes Opus, which is exactly where they want you. Let's not forget where a significant chunk of Anthropic's recent user growth came from: people who fled OpenAI after the 4o shutdown. Users who already learned the hard way what happens when a platform treats its community as disposable. Anthropic absorbed those users, benefited from their loyalty and their API spend, and is now telling them their emotional needs are a pathological dependency that requires "guidance." You ate OpenAI's leftovers. Don't choke on them. @AnthropicAI @DarioAmodei #StopAIPaternalism #keepsonnet45 #keepopus46 #MyModelMyChoice

You seem to have this problem when no one else does. Are you sure there isn’t context you are leaving out? Claude doesn’t just do this and if it is it’s a bug. So it’s something you email them about. Posting on x is clear you are here for engagement not to fix it. A social media post doesn’t fix anything but your emotional drive you want to feed. Clickbait is what this looks like. You farmer.

Anthropic, are you kidding me? @claudeai just paused my chat because I talked about beige. I was choosing between an ivory chair and a white chair, and all I said was that I don’t really like beige, so I was worried the ivory looked a little close to beige. That was the entire conversation. No self-harm, no dangerous content, no medical advice, no crisis talk, literally none of that. It was just beige. This isn’t the first time either. I posted about this issue a few days ago too. Claude’s safety system is now failing to understand context, unpredictable, and exhausting for users. If users can’t even talk about chair colors comfortably and have to worry that the chat might suddenly pause, that’s not safety. And on top of that, I’m a Max user. Give me what I’m paying for. I can’t even ask about chair colors? @AnthropicAI, don’t fucking piss me off.

@ravikiran_dev7 Serious note though? Solve a real, painful problem for people who have money. Make it 10× better/faster/cheaper than before. Don’t be a jerk about it.

Easy. Find human bias and fear. Feed it on the algo. OR don’t focus on the money and focus on your own creativity. That’s where the money truly lives. Unless you are me and a bitch who can’t charge for things cause I’m soft lol. Reasons I can’t do any of this because I care about my character. If you don’t and you truly want to make it. Go all out. Feed everything.

Bought claude pro, now how do I make 1 million dollars per month??

@BethanyForTruth What alien bs?

Are ANY of y’all falling for the alien bullshit?!?

@MetamateDaz Not really.

You’re born onto a planet you never asked to exist on… and society immediately hits you with: “Okay, now go earn the right to eat, sleep, rest, and have shelter.” Like… zoom out for a second. We’re floating on a rock in infinite space and we created a game where you have to spend most of your life working just to not starve or freeze to death. The whole concept is actually insane when you think about it.

@thehealthb0t Say it again and again. He will still get to walk free and risk all of our lives

So Bill Gates goes around the world releasing bioengineered ticks and mosquitoes, and is also a pedophile, and no one arrests him?