# created ```dataview LIST WITHOUT ID file.link FROM -"windsurf" WHERE file.cday = date(this.file.name) AND !startswith(file.folder, "windsurf") SORT file.cday ASC ``` # modified ```dataview LIST WITHOUT ID file.link FROM -"windsurf" WHERE file.mday = date(this.file.name) AND !startswith(file.folder, "windsurf") SORT file.mday ASC ``` --- # [[diary and thoughts]] #thought #diary --- # Notes Task: Turn a raw conversation transcript into a clean, fully attributed script. the Draft being discussed in the conversation is attached for context Inputs: - Transcript: raw text copied in below - Draft: attached docx including comments ) Inputs: - Transcript: [** Can you see what I'm writing? Yeah. Yeah. I can see your cursor. Yep. This is no sorry, sorry, yes. Yeah this is. This is just This is just cosmetic. Before seeing some example, always you mean okay? Yeah. Now I see. Yeah, yeah, fine. Fine fine. Yeah, this is super minor. Yeah. And uh, I was thinking then it's explaining better later. But, um, Uh, maybe we'd cross. The then I, I was wondering whether to just put that just to remove that. Uh, because the skier performance app is not clear, what which road is meant to to have, and then then it's explain it later. But here [[the analogy]] for us, it seems more just [[the analogy]] between agents and True. And we also said the last time we spoke about this that probably pseudoagent and participant as if participants should be collapsed into Position. Or even fiction. Sorry fictionalism and the yeah. That that's again. Yeah. We discussed that the other the other time. Yeah. Uh, yeah. Well another before going, that's the, the main, uh, concern about this session here. I was just thinking out in maybe something like the, the text to the image. Just to make the reading easier. And yeah, because Indeed, even because now we have [[this distinction]] between proper oriented make believe and content-oriented May believe So in a sense, we have three three versions. I was wondering where maybe it's not really the same thing, I don't know. It's there is something about that in. There's a lot of literature on fiction ideas and uh, but what, maybe prop oriented make believe is closer to to. But [[what you]] could the second one, the uh pseudo agent uh Because it's prop oriented. So in a sense it is less. It's less feature now, whereas Maybe fictionalism is more content-oriented. But yeah. For for the moment it's because you you use the the asif expression also in the in in fictionalism. So it's very out of spot. Uh, why this should be cast as different uh, views you're right? Yeah, there's too much. It's it's bringing in too many details. We should never return to anyway. Okay. Because with, when we sort of criticise [[this approach]], it applies pretty much to all of them at the same time. Testing split between literalism and not literalism. And then maybe we may say that. Yeah, there are Yeah, I don't know. I I don't know enough about this debate if Instrumental is maybe the second position is more like instrumentalism. Do you mean the zero agents one or the pseudo agent? Yeah, just like people who say, oh, for instance, there are no, electrons, electrons are just an instrument for making uh predictions. Uh, okay, first is this? Oh sorry. Now, yeah. Now I [[know what]] you mean in terms of physics. Yeah, yeah, exactly. Philosophy of Science in metaphysics in which say us a, certain scientific policies are not real. They're just useful tools for predictions. Yeah. But uh, and I don't know, I know that there is a debate, Mark, Sands a book on that other they. But I I I'm not sure whether whether there is whether instrumentalism, Mentalist without being fiction, is that? That would justify our second position. Maybe you can ask chargupt. Yeah. If he has ideas about that. Yeah. Maybe in that case, the second position can be called, probably better called something like instrumentalist, which, I think was probably the previous, or already or your pragmatists, something like that. Yeah, yeah. Maybe that. That's a, a slight difference. Because one may be pragmatic without being fictional. Mummy say, oh, we don't need the imagination. We just have tools and we can use these tools knowing that these are tools and not real things. There are artefacts in a sentence. We have bizarre defaults. On to make believe. So probably we may rely on [[this distinction]]. But yeah, but it should be better to frame it more clearly because as it stands it is Is out to see why they are different. Yeah, no, I I think there should be a reasonably easy fix to get this section. In a better shape. I think the researching should be not too hard. Yeah, he's also not crucial for the arguement exactly the arguement later. We just need to show. We argue against the the agent view as a whole we belong. Then there is some passage in which the real fictionalism is And, Yeah, then the aesthetical persons maybe even on the paper that I sent you the other time, if it's worth to considing. And what was that? Sorry, I forgot. Which one, that was. Sorry. I forgot to also but I ask I'm looking on the WhatsApp now. The morality and Aesthetics of personal Beauty. This one. Who are [[the author]] Friedel? Friedel David. Friedel and Madeline. Yeah, exactly. Is that one? Yeah. I'll write this in the comments now. Yeah. But this is also just just to add references. Yeah, yeah. Then it is the discussion of cross. I was thinking, do we? Is can we use even though you just care about art AI art? Whether we can start probably a more General model so saying, oh we can appreciate interaction even if it's not our, even if the user or is is not an artist. We can nevertheless appreciate just like we can appreciate that. The aesthetic. A merit of a conversation. Yeah, yes, it doesn't get something to go through the articles are artists, but maybe we can say, oh, even dog, he's focuses on artists. [[This model]] can be generalised to as a model to aesthetically. Appreciate uh, Or even when the by by focusing on the interaction with the user. If the user is an artist then the interaction probably is even more rewarding but in principle that the cross model can be also applied to [[Everyday Aesthetics]] good? Yeah. Yeah. I see it's maybe I'm going to put this analogy in the paper but it's almost like, uh, having a Tamagotchi or even a real pet. Now, there's sort of a It's a very good example of experiential artefact, you, you say that as though you've thought about this before, Interesting. But yeah that means to me that would be because you're sort of you're kind of prompting the the Tamagotchi to do something right to sort of make it dance around or whatever. Yeah, exactly. Okay. So yeah, that seems something to work it then. At the certain point we have say it's always a discussion of almost at the end of [[section two]], we have [[this distinction]] between participant and interlocutor. Seem to see how these ideas can be applied to the LM and appreciation, rather than trading AI as a participant, which is the cross model. We met them on a more Level Playing Field as interlocutor What's exactly the distinction. I would have thought. Yeah to me interlocate it sounds like two equals. No. Whether you're standing. New in a conversation. Yes. And I can understand that. But why not, why the cross model is not like that because he's also in a sense he's having a conversation with The yeah, I yeah, okay. I see [[what you]] mean. Cuz it's just Yeah, I just wanted some way to sort of Show that there is a difference here. It's some sort of. Yeah, the stasis or balance. What the difference is. One or maybe the the difference is is that adding it's just that cross focus on artists. Whereas so maybe participant if we consider art Just like in participatoriality, which you have the artist and then, Ah, the participants Oh, as opposed to to conversation, but it's also true that in that kind of art, the artists want to be on a par with the other people who, who participate in the performance. I but maybe it's not that, it's maybe it's something different. But I cannot tell, What is exactly the difference, okay? I mean, Uh, a moral, right? Well, is this a radical way of redoing this section? It would be to just to start with cross. And say, well, he he presents it like this and then just suggests Two other ways that it could be elaborated on. So, rather than say it's different across There's rather than participant versus personality, it's just sort of All extensions are the same idea, so maybe [[the idea]]. Is that the first route is more relational oriented? So we appreciate the interaction, the kind of exchange, whereas the the other the the second rule is more Sagittariant. So even though it's it's still a matter of interaction because a conversation remains an interaction, we focus more on the personality of of the interlocked rather than on the Yeah, the the actor within the exchange or something like that. Yeah, so maybe there's more the difference rather than So it's it's not participant versus interlocutor because also the participant is an interlocutor in a sense. Yeah, it's more something like Um, Focusing on the the exchange as opposed to focusing on on the on the other subject. Got you that's that's good, actually. Yeah. Yeah, exactly. Maybe it's already there. Maybe it's just it's just worth to say and on the first approach, the evaluative Focus, um, Shift towards what the duo does together. So it's more about the interaction itself. In the second, the interaction in just a means to appreciate the ai's personality. I'm recording you by the way. So I'll uh, because it's To always write us on the other computer on the Google Drive. So yeah. It in this way. Yeah, it's actually it's okay. I was just I'm just concern about Um, This notion of history of production, then you explained, that, that also nature. I was just wondering whether the war, whether we can find a better world than production. Production seems somehow uh, entropic as a world. You're right. Okay. So why nature products something? Yeah, I think Carlson actually uses it, but you're right, but it's not clear and I wouldn't uh, yeah. But I don't know. He's Genetic history, but genetics seems more something like biologists than um history of making. Again is philanthropic. Yeah, so even just fixed by Natural History. Yeah, but in that case is to is small is Danish or just by history, it's history experiments of nature and that by history again, it sounds entropic and By by by yeah, I'm gonna have a question. Maybe historical development, it's good. Let me just note that I'm also going to go to that page as well and see what Carlson actually says. But then then the explanation natural history of production just that at that point, it seems sounds a bit weird. Yeah. Yeah, I agree because not only does it think of Uh, mechanisms and factories, it also makes you think of Marxism or something, right? The history of production. Sounds like the means of reduction. So it's yeah, exactly seems to too entropic. Yeah. All to say, I don't know. Yeah. That that Evolution. Evolutionary history is a history of production of species like necessities, but Yeah. But I mean some when you hear people speak in those terms it's clearly a metaphor though. History has produced this or the the evolution has produced this beautiful bird you might say but it's clearly anthropomorphizing it. Just found the quotes connected to it, so I'm just gonna I'm not gonna put this in the text but that's a good marker about it. Then the next paragraph is for art. The kind is a work of art. Can you can you see that? Yeah, yeah. Yeah, that's uh why the kind it's strange to say that? The the work of art is a kind, the work of art. Usually is not a kind. It's a particular, I think the idea here is The kind is a work of art so it should actually be something like the full art. The kind Works of art in italics if you see what I mean, isn't I? I see what you're saying. Yeah, but maybe the, the problem also, also, in the previous paragraphs for nature, the kind is natural environment. Do you think just Well, I don't Do you think we could get rid of kind talk and phrase it in some other way? Because kind yeah natural environment is a kind words, it's a kind of thing. And then you have the nitrous environment in general, the natural environment in Brazil and so on and so forth. So it makes sense, but For our, especially with the, the, the, the article, a work of art. I really don't. Don't understand. What does it mean? Because a kind it's something General worker fight is something particular. Okay, so what I think this is possibly llm text, but what I think is going on is this is what it was going for. But even this is not very good, hang on, let me just turn off editing. So, I think I'll show you what. I think the intended meaning is It's something like this. So, I've just written that on the text. See what I mean? The same paragraph. So it's I'm flashing it up. Yeah, I I was lucky. Okay. Okay. Sorry. How was the last thing? Yeah. Quran the characters out. Okay, maybe that's better. Yeah, maybe I think I'm gonna double check this later because I'm not sure I like that phrasing. Anyway. Yeah, because I I'm still not sure what does it mean? Yeah, it's not very yeah. But also also, In which sense the kind is fixed by natural history of production. Because a kind and I'm not, I don't know, kind is a cut for instance, that's, that's a kind. Well, it's not an environment so a forest Forest is a kind and it's what does it mean that the forest is fixed by naturally history of production? That Forests have a specific history forest that cannot Yeah, no fair point is the same. But in the case of the artwork, it's hard to see in which sense the the kind artwork is fixed by by what? It seems more. The the particular artwork that is fixed by history product. Well, maybe the it, it means that That. Heart or an outward should be made by an artist cannot be made by bye. Bye by chance. So, but it's, it's really not, not, not not easy to understand. It's a sentence that a referee may say, just say, I cannot understand that. Yeah, fair point. I I think some of the phrasing in this section should be Tidied up. Because, One way to to make this more intelligible to me is still using Focus instead of kind Okay, so you say The focus of appreciation, is a work of art fixed by our category and history of production, but not really fixed. But I would say more something like, uh, Work of art in relation to our category and history of production. Or Yeah, and the simple nature, I would say, the focus is not rather environment in relation to natural history of production. This seems to be more more, more meaningful. Because kind for me is closer to our categories. So the the category, the kind, the category, the painting, the symphony, these are other kinds, and then yeah, artwork is a sort of, super kind. What, what seems interesting is to individually the focus of appreciation rather than the kind. But maybe maybe I'm missing something. No, I didn't. I don't think you are. I I wasn't. In the next section. I'm quite happy with the ideas but I don't like it the way it's phrased either and this section there's still some some language stuff and some clarity stuff. And yeah, I see what you mean about kind Uh, then a question about why Carson says that the artefacts are a middle domain between nature and art. Um, this is their functions. Yeah, I think that's what it is. Hang on. This is as it stands is not clear. And I guess no real reason to put it in either. Um, He does describe it as a middle domain or something in the middle or between them, but I don't really see. Actually, why that would be useful for us to Laid them out in terms of three things on a Continuum rather than just three separate things. Yeah, or or alternately one may say that there are just two domains the nature and culture and art and artefacts are both part of the of the cultural domain in the sense that our book human made as opposed to Nature, then sure. There are the the natural facts, the the but in principle artworks and works of art seems to have much more in common than nature and and then so it's the the artefacts don't seems to be in the middle since to be closer to us rather than in the middle. Man, maybe maybe we can just say artefacts Even though unless we we can find it in his book a reason why he says that? Yeah otherwise we can just say that there is our third domain beside that and nature Yeah, but You said there was You were suggesting actually, we might just might as well just do a, a two-way divide though culture versus nature. Oh yeah, that's my idea. I think there is not so much difference between artworks and and artefacts. Not sure our purpose is here now. But, On the other hand, maybe because the criteria of appreciation, Functions or maybe us maybe as in mind that in in even though in nature is not clear how much fun because the function is important in biology but in physics functional, you know, so maybe as in mind this idea that Is closer to to Nature than not because in technology function are important, then functions are important. That those in nature, especially in biology Whereas art is according to to traditional views, is functionless. Maybe, as in mind something like this. True. And again, I think it's more of a question of What we can avoid saying to avoid adding any extra complications. There is another domain. Yeah. View according to which art requires a disinterested up so it's free. Beauty art is free Beauty whereas nature and artefacts are. Yeah dependant. Yeah, functional Beauty. Yeah, yeah. Um It also just occurred to me as well as if we do it. Something like this. It might actually also help us avoid this history of production term completely. And even the history is not that important for our purposes. It's the the realisation. So if you, if you think about. Yeah, so, I mean, we both know this this literature but it's it's more about the process is more our sort of uh, process product. That is the idea that you don't have just to consider what show up, but rather the process that has led to that. So, it's, it's another natural distinction. Uh, exactly. But also fits in very well What I mean specifically is, is about the The reason I'm asking about this or mentioning this is because, um, This will get towards the. Sorry, beg your pardon. When we talk about the Aesthetics of llms at the end, What the idea I'm still trying to get at is to do with Mechanisms being constituted by all of text culture. Okay, that's you see what I mean? So, that's the, yeah, I think what you have to develop more. Uh, this idea it's it's a bit like culture, uh, the, the young training, set, or data set the, it displays the role of matter in uh, in in uh, environmental static. So the idea that we can appreciate uh, mounting also for the kind of rock the Earth. So in a sense, Prince of geology, it takes physics as a smarter and then adds a configurations that are typical of geology and generative, AI takes culture all the, the, the, the website, and the booths, and then shape that in a way, uh, good. Yes, the way in which river shape the matter of Was in some previous version that I think we should. Bring back at a certain point. This yeah, he said, we in a sense, we see culture through technology basically that that's the the it's, uh, it's an environment, which is made of, of cultural element. But organised within this, this, uh, Simeon semiotic physics. Uh, Exactly. So yeah I think that that that's that's important. Yeah yeah so yeah I think I think the way you're suggesting for this section I think I can reorganise it in a way that better lays the groundwork. Later on. And I think it's a lot about cutting out some of this terminology as well. So, Yeah, I I think the notion of artifactual environment that you show up for the first time, uh, in the last paragraph of section three. Oh, he does. Yeah. Oh, it shouldn't then, sorry. Yeah, that must be copied in from. Yeah. Trying in section 4, in particular, I've been really Sorry what I mean is in section three actually, it's been so hard to talk about Carlson without getting into the details of what environments are yet. And so yeah, this definitely shouldn't be. Bad Section 4. I think it's going to get just there's some Notions that that still require more more Verification for an average philosophy reader, the very notion of talking I think in a previous version there was a a definition of what a talking in the fact that it can be a world but also, but it's no longer that it disappeared. So I I think it would be worth. Then I find a bit obscure. Also the third paragraph. Three lines of the third paragraph secondary dial which is the use case layer by alignment and interface some LLM. Deployments never present dialogue some non-nlm system, do that like the four cannot fix kind learnt continuation. Can okay. I sorry, I have a comment on the doc, but I, I was going through this before a conversation and I have exactly the same comment on the docu. It's very, very bad and unclear, but I I just need to be expanded, sorry, carrier They have an intuition or what can mean, baptists. I think they require more. Yeah, more clarification. This is actually a slight worry as I elaborate here. So, In this the bit which you highlighted was this is the paragraph you're talking about. So just the last two lines. When it's talking about alignments. That's about post training. And um, post training is when it stops being just a fire hose of text and starts being uh, I mean to look at it. Okay, it starts. So after all of the training with all the corpora, it still doesn't function like a chatbot and you have to have all this extra trading on top of it, to make it function like a like a person. And so the difficulty here, Is going to be getting enough information there. So we're fully explaining what is happening without getting too bogged down in all the details about. Yeah, exactly. Yeah. So yeah this has to do probably with this the fan for instance alignment interface, it's intuitive what it is but alignment is another notion that is used without being defined that is not easy to to to understand post training alignment. Yeah, it's it seems a very important notion but without a proper characterisation is hard to understand interface convention that's more. There is an interface and Convention just the rules. Uh, Sounds very mysterious. It's very jargony. Yeah, yeah. Unless it's one way of that, in that case probably is just. So it's a farther layer of rules. A post that are that are established post training, so, for instance, If the prompt is about paedophilia don't don't reply. I cannot say that or if you in in the in your training, you end up on copyrighted material just reply some sort of normative layer that that filters in a sense that selected the Yeah, does it? I mean, that's true to litre. Well, I don't know. I mean, you made me think maybe there's an easy way of getting into this, which is the way people sum up, alignment, or post trading, as they say, it's to make the llm. Less honest and helpful. This is the three ages basically. And if you think, maybe if I use a head puller which is the third one honest, Honest. Yeah. Um and maybe that's a way of leading in with that. Yeah, that's the goal I was thinking. Also what's, what's the means to this end? The means it doesn't seems to be I don't know, it's still neural network or is sort of it's another way it's still working on neural networks. So it's another training in a sense. Yeah yeah. And like part of the training for example involves Texts of dialogues being fed in where one of the dialogues is. The interlocutors is. A helpful assistant. You see? And so it starts training for predicting the next thing in a helpful assistant conversation and that feeds into how it gets to the rest of the, the system. Because then there is similar problems in the next paragraphic, which is again another time to, by state. I mean, by operator, I mean that's quite helpful. It's just at first. Uh, do we really need the the the the mathematical formula I have seen? I have another comment here saying we don't need the mathematical formula. It's on, sorry. Continuation. The maps are state to a distribution of the next tokens. I think that's enough police. It's okay, constraints, it's a bit obscure. I mean transformational masking of the distribution, prior to selection. For example, alignment and safety filter system prompts What else the next beta is much more. Clear, any gate in that restricts selectable outputs? I wonder whether we can just say that this is clear. So constraints are gatings that restrict selectable Apple. So the the model uh the basic model uh just talking prediction and you just predicts and predicts. Yeah. If we predict something which is uh violence copyright it prediction. Nevertheless he predict something which is uh uh as you say dishonest or an helpful whereas with these constraints there are there is gating and so he's forced not to predict certain things and and this future to predict others and I wonder whether this is enough to to clarify this notion of alignment, if you can just say that because these are the terms like transformation masking of the distribution, these are two dragons here. I mean even if it's too simple just to say gating that's maybe a good way to start. Yeah. These are all gay. This is also the prompts in a sense. It's a way of constraining. The the prediction even though the the role of the processing is different because seems more an input. So it's not just that it's what start that the predictive process rather than oh the system prompt. You mean yeah the system ground Quite yeah, it's not the same because yeah, even System. Plumps are not part of post training. They are right in this sentence. Yeah, they shouldn't be. It shouldn't be there because their their post post training, you do the post trading then that you start the conversation. Yeah. Yeah. Anyway, that that's, I think are not too difficult to fix. No, again, all of this so far I think is not Especially. Um, by the paragraph that starts with a more pointed version of the objection. Badly phrased. Sorry. I say it. Yeah, yeah. Yeah. Wait what was it at that point, he's not clear, what's good? I think I would say something like a more pointed defence of the ancient view. I think that that's what you mean, right? In the same bit later. When, as a final objection. I would say a final defence of the agent view. Yeah. Then the previous paragraph there is a sentence that again I find obscure theoryology enters, only as a constrained finished state of the micro time scale of generation. The process. Halts when the coding policy, selects a token and the loop advances. There is no diagonic game spine turns Is not discovered interesting. Cancer of the artefacts. I I can more or less see what what? And then the conclusion uh about cartoon is it? It's okay but this paragraph I think require again clarification and yeah but you the paragraph is pretty shit I think we can say. Yeah. So just send a message to being changed. So maybe, okay. Uh, yeah, because I, I told him to meet that half past five, but I will, I say that I will call him later. Okay. Okay, then The the transition just after the the paragraph where the last paragraph because the representation in question are tokenized it it's okay. It's just a bra to abrupt. I I I think we need something like After seeing that the Argent view doesn't work, we we focus on the semiatric alternatives and then start introducing that because Now it's a block, yeah. Yeah, I couldn't note then now But yeah, section five. I like it a lot. I like the night I wonder whether this can whether we really need a section of its own or this can just be Used from the beginning instead of having Uh production history things we can just say, oh, the way in which we we frame a Carson count is by using this distinction. I don't know. It also was well as it is, it's just that, that it might make the previous section. Look clear. Then I don't know. Yeah. Unless we find a way to to, to be, I don't know. It's supposed to work. I like it and maybe. As we introduced this notion of narculus and naturalata at the at the beginning of the second paragraph and then they show up again at the end of the third paragraph. Maybe recalling them in the middle would help. Just not to, to just say, oh what this is what we call now to us. This is called because it's a long way from there are like dirty lines and the risk is that the reader doesn't say they come as a pair. Yeah. Yeah. One more follow-up to what you just said about. Where we should put some of this information. Maybe it's it's because then also I like a lot. Also this this definition of generative environment, so it's also more specific of us, so maybe it's, it's better just to introduce cars on uh, with his own terms and just maybe signalling when where, where something is not clear trying to make it more clear. And then here we start saying, oh, but we prefer to frame, uh, his proposal by using this this distinction that we introduced. Yeah, yeah. Okay, then section 6, I think is, is the key. And and I think it's the, the idea, very good. I think we need more again more rhetoric. Uh, Effectiveness because here the idea I think is something like okay, the agent Works not functioning And but, so what's the? The focus of appreciation. It's it's not in a sense. So the artefact is the right the right way to go. It's not nature, it's not an agent. It's not a work of art, we don't even have to to demonstrate that. Nobody is saying that that GPS will work far. But in a sense is not also an artefact that this is not an artefact in the in the in the strict sense because it's not like appreciating a hammer or that's the tool uh the tool model that we challenged in the other. Because it seems that they it would be weird to say oh for for for um a washing machine, the focus of appreciation for the aesthetic of design of washing machine, is the was the English. Then the bucato in Italian is the This episode in which you, you put your laundry into the washing machine and then you take it clean. Not the focus, the focus of design appreciation is not that it's not the laundry. Uh whereas in this case it seems that it's that. So there seems to be a difference that or maybe we want to focus on that instead of from the Arctic even though we can also select the artefact cannot. But I'm not quite what follow you can. I just double check. Yeah I can understand because that's that's for me the most trick but This was a split that's actually been in a previous version when we had that quote from the guy from anthropic about growing llms, remember that, quote? Sorry. Yes, yes. Um, and basically that's I think what you're saying is The realisation of function in an artefact, it seems maybe the difference between ordinary artefacts and llms is maybe not the real age. Well, the realisation is different. It's built in a completely different way. Is that what you're getting at the sort of? Yeah, but also, there's something. But the other thing is that we say we have this notion of artifactual environments use user results, Now I know you're getting it and so we're not appreciating the laws of nature of the laws of physics in extractor. Ok, we are we are appreciating what they determine what they are. play music of mixture between Thanks for the systems in question are made from The Motorbike culture makes them because I'm in the introduction especially we need to find things much more rhetorically into the settings up on the same section basically. Play radio the teaching of technologies that turn on the TV DVD for me. Yes, just bring it. To connect with the data of the good. Yeah, that's one more just talk about with your physics paper the package in Paris you made me think that one is here as well is so when you talk about a model that the actual model produced at the end of the system that the model is that the model ok? That's normal equation which is derived from all the training centre. Just a huge long equation. This is able to appreciate this particular level right in the same way you're talking about potentially appreciating physics strapped as well good for you. Scientific knowledge as well actually. Ok, would you put data from 30 to 20? The end of 7 do you have more to say about 5 p.m. For the Asian today if you want to look at the features of the process for selection? That's good, so yeah, I think maybe something to think about it. Just leaving section every day is Easter good and also something else with maybe we should mention as well this again to be fair to him. He talks about exploration of latent space as well which is very similar to environmental stuff in the future would be comparing that the exploration paradigm generative part. doesn't Feature of making dessert place under the age of the tallaght self to do it to make other stuff to tell something very strange. Far far to mention this because something going on something is contractors in the ontology of what is an agencies in the making. Play moana on the paragraph giving that Father very descriptive methodological. Four sections 8 and 9 if you want I mean do you even want to talk about these in such a bad state right now? It might be better to I need to I'm so close. This is the if I can just get the last two sections Everything's Gonna Come Together I think so. Yeah, I agree. That is very Avalanche explained. 222 yeah What time does today's what Friday tomorrow night's work on something else for the customer reviews I've got a right about propaganda for this frogs have a deadline at the end of the month so I need to do a little bit of that but anyway the to you asking for my email address from university invited in November to give a talk about AI and propaganda propaganda. I think I've seen one he's got a couple now. play Reasons to be Cheerful all of the saying about this. I'm finding it so I don't know anything about propaganda on my phone. paper that if you tell me the title of the Yeah Yeah Yeahs She's not the only one in the UK and the recommended getting in touch with her university. Is it good? I don't know just seems to be a bit of a dick head. You can stop the radio, please. Thank you for that reference in Stanley in Ross on Sunday what I think. I'll do some work solely on satellite and section 9 and maybe if you have time sometime next week. We can talk about specifically that idea after the conference before we possibly what time what day are you going into? And by the way to set up the company from general that you can go there is not going to be before radio on an otherwise, what if what if the trains are late and we miss half a conference. You know could you tell me if it does help you? Everybody of the ability to go to the conference without waking up in the night. Dad work, let's so he said he pencil in Monday afternoon and then I'll talk to you will see what day is the class of the conference was to work on the on the paper and you prefer probably I'll go in quite late. Especially general probably won't pay for my dinner insurance. Monday I can do anytime volume OK meeting at 3 on Monday on before you got a Monday later. OK that is also have you already know how to send a night 90% even just a group I presented with flights on this topic. Generally very good. What helps a lot as well is make slides out of this copy slides for the format knows what you want to repeat so if you would like a bullet points and sub bullet points slight it seems that you probably need a week things and little bits and pieces but I know it should work very easily something useful also sorry. This is only the text the slides. I don't think I think it's placed into the PowerPoint that is the last time I don't. Yes, can you produce one more thing if you're thinking of using it for slide stuff as well if there's an option called canvas. Have you used as yet? I used to have working on that you have written in text and it's not quite right. So you delete a little bit yourself or you say easier to text to work on ok? Let's stay in touch. Can you send my regards to vincenzo? **] - Context (optional): [paste any referenced documents or excerpts here] Requirements: - Preserve the entire conversation; do not summarize or omit content. Keep the original language. - Attribute every utterance to a specific speaker. Use names found in the transcript or Context when available; otherwise assign “Speaker A”, “Speaker B”, etc., and keep them consistent throughout. - Correct punctuation, casing, spacing, and obvious transcription errors. Lightly edit for readability without changing meaning. - Remove non-meaningful fillers (e.g., “um”, “uh”) unless they affect tone or meaning. Keep meaningful cues like [laughter], [sigh], [crosstalk]. - Keep timestamps if present; retain them where they appear (e.g., [hh:mm:ss]). - For unclear audio, use [inaudible], [unintelligible], or a best-guess tag like [term?]. - Use the Context only to disambiguate names, acronyms, and document references that speakers mention. Do not add facts not spoken. If a reference is ambiguous across multiple documents, keep the speaker’s wording without forcing a choice. - When a speaker reads from a document, keep that content in quotes within their turn. Output format (no extra commentary): Speaker Name: text of the utterance... Speaker Name: next utterance... One turn per paragraph with a blank line between turns. Return only the formatted transcript. DRAFT: (attached) Requirements: - Preserve the entire conversation; do not summarize or omit content. Keep the original language. - Attribute every utterance to a specific speaker. Use names found in the transcript or Context when available; otherwise assign “Speaker A”, “Speaker B”, etc., and keep them consistent throughout. - Correct punctuation, casing, spacing, and obvious transcription errors. Lightly edit for readability without changing meaning. - Remove non-meaningful fillers (e.g., “um”, “uh”) unless they affect tone or meaning. Keep meaningful cues like [laughter], [sigh], [crosstalk]. - Keep timestamps if present; retain them where they appear (e.g., [hh:mm:ss]). - For unclear audio, use [inaudible], [unintelligible], or a best-guess tag like [term?]. - Use the Context only to disambiguate names, acronyms, and document references that speakers mention. Do not add facts not spoken. If a reference is ambiguous across multiple documents, keep the speaker’s wording without forcing a choice. - When a speaker reads from a document, keep that content in quotes within their turn. Output format (no extra commentary): Speaker Name: text of the utterance... Speaker Name: next utterance... One turn per paragraph with a blank line between turns. Return only the formatted transcript.