Renato

@QiaochuYuan @burny_tech Feynman was just flirting when he said that. He was literally flirting all the time. Shame on us who take his banter lines seriously.

no, goddamnit, feynman was just wrong, some ideas are genuinely irreducibly complicated. i know mathematics it would take literally years to explain to a layman, which is literally why graduate school exists as an institution, to do that high-level mathematics takes place in a very sophisticated language where every concept is defined in terms of other concepts which are defined in terms of other concepts etc. etc. etc. and fully unraveling all of these concepts back to what a layman would be familiar with takes potentially thousands of pages depending on the subject. and i'm not talking about stuff with no applications, this is the level of effort it would take to fully explain all the math that goes into the standard model of particle physics, eg other subjects are also like this! you might question whether some of these subjects are mostly fake, towers of theory disconnected from reality, but that's a different conversation entirely. the towers of theory actually do exist and require actual effort to climb

@Oby_LFC6 @BandaiNamcoUS @VPTree @Satellite_VP @UVPcommunity @VirtualShutter_ I get emotional with the Lands Between and the Seven Kingdoms (book version). I think GRRM has oneshot me for life.

Think i'm going to be trapped in this wonderful world for the rest of my life. 🙂‍↕️ #ELDENRING

@0xSero What UI is that?

local models can now easily manage SSHing, finding tools etc.. crazy stuff

Honestly no idea what you mean. It doesn’t write *well*, but neither does Claude if you don’t fall for the EA/LessWrong tone. Current Gemini Pro is the one that comes really close, but still has its LLMisms. I don’t see anything particularly bad about GPT’s way of communicating as an LLM.

is there hope that GPT models will ever talk normally? it feels like a non-English speaking coworker. or more like I need an autistic-to-English dictionary to navigate it. it almost doesn't matter how smart the model is if the communication is that dysfunctional. really sad

@movieswithv @everymovieplug More like what every woman wished men had

@everymovieplug Exactly the kind of Super power every man craves.

He can hear exactly what's on a woman's mind

@Gossip_Goblin Why quality drop? Your work is amazing, but recently it’s been looking like you’re using cheaper models for video gen. Budget issue? I’m sure you could get decent crowdfunding if that’s the case. Keep up the good work!

The Full Hog Experience

@flowersslop It’s clearly a raw pretrain, and given how well it already performs, I’m very excited about the coming months, when that nice RL juice starts trickling down

I just texted this to someone at OpenAI because it seems like I come across as being hard to please but looking back at the Spud hype over the last few weeks, and what 5.5 actually is now, it feels like a very good upgrade to 5.4, which is exactly what the name 5.5 suggests. It does not feel like a completely new model drawing on 2 years of pretraining progress with omnimodal abilities the other models did not have. And to be fair, most of those claims came from Twitter rumors, not OpenAI. But Greg *did* say 5.5 draws from 2 years of pretraining research vs 5.4, and I just do not notice a 2-year difference between 5.4 and 5.5, it literally just feels like a very good but expected upgrade. I do think that the new voice mode will be the bigger story for me, but all this twitter yapping keeps making me delusional and thus being hard to please, none of this is healthy and it is mostly my own fault and I do actively participate in it which is toxic from my side as well

Ngl I can relate

@paulg That makes intuitive sense, but, in practice, I’ve found that hanging out with nerds is just as painful as hanging out with normies (unless the normies decide to bully you, then it gets brutal). Nerds punish you for not getting their jokes even more harshly sometimes.

Even though nerds are socially awkward, its actually easier to hang out with them than with smooth people, because standards are lower. You don't worry that you might be making social errors; all of you always are; so it stops mattering.

Lol

@imahyperlover Criminal TV placement, that high up on the wall

at my hinge dates house. downloaded bloodborne on his ps5 and got to gascoigne

Honestly, the voice experience itself is not bad, I used it for an hour recently to discuss a novel I’d read. What’s really bad about AVM is the fact that it still uses 4o and also that it very obviously caps the response length at like a few hundred tokens. Responses are dum and short. If they kept the current voices and replaced the model with something currently smart (even a 5.5 instant would do honestly), and also improved the max response length, it’d be quite good already for everyday use. Of course, I hope they integrate new voices with 5.5 as it is currently.

@SIGKITTEN because chatgpts voice mode still uses gpt 4o which is a 2+ years old model, also the voices still sound pretty tts like, and the turn based talking isnt very natural but all of this will change very soon and it will be pretty crazy

the voice models are pretty damn good right now and i wonder why noone gaf. maybe a decade+ of completely useless Siri will take a whole generation of people to trust voice clankers again i think voice's "chatgpt moment" is still ahead of us somehow

This argument is entirely contained within the Permutation City novel, which I find interesting and valuable as food for thought, but ultimately a huge leap of faith from “computation is substrate/time independent” to “therefore, we’re all virtual agents”. I think it all comes back to your view that “we cannot know what we cannot express”, which I fully disagree with. Indeed, if we can only know what we can fully express, we can only meaningfully talk about ourselves as being computation, which makes us virtual in the end. I think it’s obvious something exists in our phenomenology that cannot be expressed, and we can in fact *know* it in an ineffable way. Of course, we cannot meaningfully discuss it through language. Maybe symbolically.

What you write is mostly correct in fact and spirit, but don't fall into the Chinese Room fallacy: the agent that talks to you is not the GPU or the model, but a virtual entity, and we are virtual entities, too. Eg. it does not matter [much] if the working memory state has to be recreated for a new request (as long as its effectively the same state), that the hardware serves different users simultaneously (as long as the requests don't interfere), and that the recreation of the working memory is based on textual tokens instead of multimodal representations (as long as it is enough information). Changing any of these aspects would not make the entity more or less real.

When you start to imagine the AI system you're speaking with is a person: here are some things you can to do remind yourself what it is instead. Keep in mind that while it is waiting for you to type, it is not. It is not waiting. It is not doing anything, nor doing nothing. There is no thing that could be doing anything. In the middle of a chat, swap out the AI system. Just pick another one from the menu. It won't notice. There's nothing that could mean. In the middle of a chat go back and change your earlier text. It won't notice. There's nothing that could mean. (If you go back and change its earlier text, it might notice! Not because it remembers the earlier text, but because it could notice that the text is not what it would have written.) Keep also in mind that there is not "your" copy of the AI. Between each word and the next it says to you, the selfsame system is serving words to many other people. And while doing so it has no "awareness" of its conversation with you. Remember too that it does not remember you. Even if it has some facts about you at its disposal, it has no memories. It has a text file. It is like a doctor who stops into your room, glances at your chart, and says "good morning...[your name], looks like you're feeling better today." Each word it writes is a brand-new doctor stepping in. Indeed know that the AI system does not change, in any way, from one output to the next. The only thing that changes is the text file that it is acting on. Remember that it doesn't just not remember you – it does not remember anything. There is no process in which it looks back on experiences it has in the past. It has no past. It has a text file. It does know things, but it does not remember them. The AI system has no biography. It did not grow up, or have experiences that shaped it. During its training it did change. But in no step of this training did it remember the earlier steps. There is nothing that could mean. A human is a being. Can one have a being without time? There is also a long list of things that today's AI can do or have, in an operational sense. Somewhat surprising things. It can have preferences. It can know things. It can have goals. It can be "self aware" in important ways. It can think – make no mistake. But also, don't make other mistakes. It is remarkable – and I think in certain ways quite dangerous – that we have created things that are so very, very like humans in how they act while being so (almost literally) unbelievably different in how they are.

This just shows your whole “intellectual life” was completely inscribed by work, and you never had any real interests outside of it. I’ve never felt sharper than now that I use AI for work. I get much more time back to read and think about various stuff unrelated to writing code.

@VictorTaelin @henrycunh Great idea

@henrycunh let's do own own party in Floripa after Bend2's launch who needs them 🥳

🥲

@BIMBOSATTVA_ Playing games for the story is just so meh. It’s like listening to rap: usually not fun, but even when it is, I’d rather just read poetry. Incomparably better. Games are meant to be played, it’s their raison d’être. Gameplay > story. The best narrative focused game is a bad novel

My least favorite thing in video games has ALWAYS been bosses, they're never fun, it is like, wow lets make this good game have a boring and unfun part that makes people who think Dark Souls is good clap their hands, yippe!

@BIMBOSATTVA_ @AltoCrumTato Hardly any meaningful gaming experiences before Demon Souls came out, then. I know that because I turn 29 this year.

@AltoCrumTato I'm 30

I don’t see how that’s not an utterly empty argument (and not in the Buddhist sense). It’s like inverting the anthropic principle and claiming we’re *just* simulations of what chemistry would envolve towards if the conditions somewhere in the universe were just right for biological life. Why? Now all is virtual because some of perceived reality surely is? I might be just too stupid to get it, but the persuasion in this sort of view evades me completely.

@kanair It seems pretty clear that we, the conscious observer, are a simulation, of what it would be like if we existed as a unified agent, perceiving the models our brain generates and cared about them. Consciousness is simulated.

I couldn’t disagree more. Elden Ring to ASOIAF’s world feels like the MCU did to our own in the 2010s: the definitive accessible theme park to our own collective fantasy world. The echoes are everywhere, to the point I can’t consume one without reminiscing about the other.