It's Just A Model

011 - It's Just A Model - McMansion

• Peet Sneekes, Ron Kersic • Season 1 • Episode 11

Share episode

Use Left/Right to seek, Home/End to jump to start or end. Hold shift to jump forward or backward.

0:00 | 1:11:03

It's Just a Model is a biweekly conversation with Ron and Peet on the interstitium between innovation, tech, design, and our human behaviour.

Your hosts are Ron Kersic and Peet Sneekes.

In this podcast, we’ll be talking about the good, the bad and the interesting, all wrapped into an informal, unscripted conversation.

CHAPTERS

  • 00:00 — Intro
  • 01:21 — The  Predictability Myth
  • 07:28 — Against the Excel-ification of Everything
  • 14:55 — 10 Incarnations of Yourself
  • 18:54 — Talking Dirty
  • 29:45 — Surprise Machines
  • 37:50 — World Models
  • 43;37 — The Ultimate Editor
  • 49:20 — AI Manifestation 
  • 56:45 — DIY All the Things
  • 1:09:15 — The Shape of Innovation

LINKS

ABOUT US

SUBSCRIBE

SPEAKER_01

Are you ready Ron?

SPEAKER_00

Ferry.

SPEAKER_01

Because this is It's Just a Model, a series of ongoing conversations on the interstitium of innovation, design, philosophy and tech with my friend Ron Kursich and myself, Pete Snakers. And yeah, let's kick it off. It's 11!

SPEAKER_00

It's the last episode that we can pretend it's binary.

SPEAKER_01

No, no, no, it's two ones. Yes. Oh yes.

SPEAKER_00

Okay, until we get to 100.

SPEAKER_01

Yeah. Yeah. Oh man. That's that's that's uh that feels a little bit sad.

SPEAKER_00

Well on that note.

SPEAKER_01

Yeah. So um oh man, I should I should probably read my rant because I I made a rant about uh perhaps you could could paraphrase your your understanding for uh of my rant because so we we we we never prepare an episode of it's just a model, yeah.

SPEAKER_00

But we drop text at a notion page, yeah, which we typically ignore.

SPEAKER_01

Well you do. Yeah, I try to remember things at this moment, it's it's like weird, it's it's a vacuum.

SPEAKER_00

Yeah. And you had this both nice and disturbingly rare trend that you uh that you're wondering why generative AI can't be predictable. Why if you I mean because you're paid and you want to have a service, and that service needs to be predictable, repeatable, understandable, all these kind of ebbles.

SPEAKER_01

Yeah, yeah. Well, it's it's kind of like a it was a speculative uh uh design kind of uh uh rant. So I was I find myself in situations that people keep asking, like, oh no, no, it starts hallucinating on uh about AI.

SPEAKER_00

I wish you did.

SPEAKER_01

And uh oh, I got some wonderful code when I was vibe coding, and then I need to do a little change, and all of a sudden it made something totally different, and it was all ruined. And it seems like we try to tame AI specifically as if it was like one of our classical products, like classical services, you know, like Word. If you select a thing and you press the fat B, you get a fat text every time, all the time, exactly the same. It's repeatable and it's really, really lovely for people that that like such products. And similarly, if I make an espresso, we just made an espresso, I have a certain amount of coffee with a certain grind, with a certain amount of water, and I get this wonderful espresso, and it's lovely, it's really lovely, and it seems like AI is living in a different realm. Yes, in a different oh, there's this wonderful word, so you use it oftentimes. Uh, paradigm.

SPEAKER_00

It's a different paradigm. I feel now like this a live action role-playing consultant, yeah, yeah, yeah. Disruption. Yeah, exactly.

SPEAKER_01

It disrupts everything, but in a bad way. And I feel like we have we have had conversations about AI as if it's no, it's not a uh something that can predict things, it's not something that will do uh give you the truth per se. No, but it's more like a muse, it's more like something that informs you, it uh helps you create.

SPEAKER_00

Yeah, and I mean it doesn't give you the truth per se, it doesn't give you the truth. Period. Yeah. Because that's not what's trained on, right? It's about it does plausible things, it generates text as if it was the answer. Yeah, exactly. You get a scenario of an answer. Yeah, you get a simulation of an answer.

SPEAKER_01

Photorealistic answer.

SPEAKER_00

Yeah, I mean, sometimes it's photorealistic. Yeah. No, I mean honestly, what it can be plausible, it can be preferable, it can be possible, and it can be totally out of rack.

SPEAKER_01

Yeah, exactly.

SPEAKER_00

As in scenario planning, right? In scenario planning, you have this scenario, this future scone. Not sure if you know that one. Yeah, definitely. Yeah, yeah. This is what you have. You in the show notes. Absolutely. Oh yeah, we're we're gonna make that. But it's the uh that's what you do, you're you're traveling through the future scone in that direction and also up and down.

SPEAKER_01

Oh, that's interesting.

SPEAKER_00

So So that's why you're that's why I think I mean that's why I I wish it did hallucinate because it in the end it it's still stuck in its own frame, right? Yeah, yeah. You never get JGPT to hallucinate.

SPEAKER_01

So if we use that cone of of uh future cone, the future cone. The future cone, so that's something like this, and you start out with the very narrow part that when they intersect, and the more you get into the future, the broader the cone gets. Yeah, i.e., the more possible futures there are.

SPEAKER_00

Yeah, I mean yeah, I mean and a greater variety of futures, right? I mean, so if you just continue, then you get to the preferable, assuming that you think that the present is preferable. Yeah, right? Yeah, yeah, yeah. There's less change. Less change. Yeah. I mean, and then so you uh and then you have the the the possible of then you the where it slightly deviates, yeah, right? It's it's still possible slash acceptable. And it gets a bit broader, it's plausible, yeah, right, and it gets broader and broader, and then you get to the to the to the wacky stuff. Proposperus, like I can't pronounce that word, so that's why I kept actually uh not not using it. And that's the thing, right? It deviates, so the the whole realm of what it can come up with gets bigger and bigger, yeah. Which is fun. I mean, there is the thing that you said never thought about it, and that's that's I mean, that's the secret sauce. Never thought about it.

SPEAKER_01

Okay, so there's two things. At this moment, we try to make the the AI cone, we try to tame it, we try to make it narrow as narrow as possible because we don't want we want predictability, or at least we use it as if it was something that needs to be predictable.

SPEAKER_00

I mean, because for like 80 years, right, we had a computer that was predictable.

SPEAKER_01

Yeah.

SPEAKER_00

A computer that was accurate.

SPEAKER_01

Yeah, right.

SPEAKER_00

Computer says yes, yes, or a computer says it says no, and if the computer says no, it's no. Exactly.

SPEAKER_01

Computer says no. Yeah, hello.

SPEAKER_00

I mean, I mean computer can't can't can't be wrong.

SPEAKER_01

Within the given re uh uh rules, it says exactly this all the time.

SPEAKER_00

Yeah, yeah. Yeah, and I mean, and but that's not what life is, right? No, I mean so I mean well if you go to a restaurant, right, your favorite restaurant, yeah. I mean the the the uh the menu is pretty much fixed, right? I mean, but even then it it changes sometimes. This is a bit bigger. Yeah, sometimes the waiter or waitress has a very good day or a very bad day, right? Sometimes you're sitting next to annoying people, sometimes you're sitting to wonderful people, okay, right?

SPEAKER_01

It's the same service as our perception also changes the product itself.

SPEAKER_00

I mean, you're actually part of something b bigger, right? So I mean, and and and I think one of the reasons why I go to my favorite restaurants is because basically the spectrum of surprises, yeah, without the uh the the the negative surprises, right? The food will be will will be decent enough and etc etc etc. But it's and I I'm so oh this this whole excelification of everything, right? I mean that everything needs to be in Excel and it and and and and every ver variance has to be eradicated because variance is bad. Yeah, yeah, yeah. Right? And I get it. If you want to do a mortgage calculation, I totally get it, right? You don't want any variance as in Don't you say aye in mortgage calculation? Absolutely not. No, not generative.

SPEAKER_01

Because we already know how to do it. It's not it's not rocket surgery.

SPEAKER_00

But there's so many so many things we we couldn't automate, and and now we can, as in it comes in a box.

SPEAKER_01

Yeah.

SPEAKER_00

I mean, just I mean the the oh god, I'm I'm s I You're smoldering. I mean, if if if somehow, right? I mean, all AI data centers would vanish from the face of the earth. Yeah, I would be in a really bad place. I mean, life without Chat GPT or Grog, yes, Groc, or Clote, or Perplexity, can't imagine it anymore.

SPEAKER_01

Oh man. I think I think you're right. And still, I still kind of like so the the cone that we just created, and people wanting to make the cone as narrow as possible. Let's say the AI possible uh the AI usability, no predictability uh cone, yeah, right? Yeah, so we make it very narrow, but at the same time, there's another factor, which is all the data that we put in that thing, that also narrows it down, right? So the results that we get, so let's say you want you want it as broad as possible because you want wonderful things, you want things that you even couldn't even think of. That means you want as broad as a cone as it can be, yeah, but still it's limited to the information that comes in that goes in there.

SPEAKER_00

Absolutely. And that's and and I mean if there is one thing that annoys me, it's that. Exactly. Right? I mean that it just how do we fix that?

SPEAKER_01

Uh listen more to people, even more, like like uh collect data and stuff like that.

SPEAKER_00

It's the I mean I mean, in in the end, the the the the the ability to to reframe the discussion or the topic at hand, it's us. Yeah, right. So must be last week read this really great article that said that uh um generative ai is very good at inductive reasoning. Yeah, deductive reasoning they also mastered.

SPEAKER_01

Yeah, yeah.

SPEAKER_00

But what they can't do is abductive reasoning. Oh right, so that you have limited facts, that you have uh observations that don't make sense, and that you come up with a hypothesis that explains it.

SPEAKER_01

Oh.

SPEAKER_00

Right? And that's what we are really good at.

SPEAKER_01

You mean like uh like uh something that also makes sense rather than dreaming up some.

SPEAKER_00

I mean that that makes sense enough that it is convincing enough to start to start reasoning within that frame.

SPEAKER_01

Yeah, yeah, yeah.

SPEAKER_00

Right? But induction and up uh and deduction, they have a the had they assume a certain logic, a certain frame. Yeah, and with abductive reasoning, you change the frame.

SPEAKER_01

Okay. Is that something like uh extra extrapolation? So you have a set of data and you start thinking up the possible data outside of the set?

SPEAKER_00

It's it's it's inference, it's inferring of an incomplete incomplete set of observations, and you come up with what's what is the bigger picture? Yeah, yeah, yeah. Right? And that's not true or wrong yet. Just let's see how fast this uh this this this gets us.

SPEAKER_01

Seems like my job, kind of.

SPEAKER_00

I was about to say that. Which is a very human job.

SPEAKER_01

It's also lovely, it's really nice. You can you can make like um, yeah, it's speculation. It's kind of like so, what what else would happen from this point on? Yeah. So what what if pigs crash into the earth somehow?

SPEAKER_00

Yeah, right? Yeah, and and the only real reason to get an LLM to uh the only way to get an LLM to reason in in that realm is by you bringing it up.

SPEAKER_01

Yeah, exactly. Because there's no data.

SPEAKER_00

I mean, yeah, the the the the there is no pattern in the generally observed training data that would give rise to this one, because otherwise it wouldn't be uh abductive reasoning anymore, because then it would be normal to either be induction or deduction, right? But this is the the basically the inverse, it's it's it's it's the unknown unknown-ish, right? Or the slightly known unknown-ish.

SPEAKER_01

Yeah, which is in I mean, if you talk about humans, it's also kind of known. The only silly thing is, I mean, we have like this this very weird way to uh ideate, right? To connect uh concepts with ideas, yeah. Yeah, which is uh which is quite interesting. I I I at the same time I'm also kind of wondering because if we have AI, and AI is kind of like the statistically truth, it it comes up with stick statistically plausible things, plausible. So what if and so if we want to make it um hallucinate intentionally, like really badly, then we should probably say, okay, rather than taking the statistically plausible answer, do the inverse of that. Yeah, g give me like the small, not so the statistically everybody would say this except for five five other people. They would say some they have said something else. Yeah, use that data. Yeah, can we make that?

SPEAKER_00

Well, you have this temperature thing, right? So that you get a greater var uh variety in the uh uh the top scores it takes. Yeah, so it's the uh yeah, uh but I'm oh that's it it's a nice one.

SPEAKER_01

I'm kind of like uh the trying to trigger that creativity and use it as as a okay. You use the complete set of the human being because the the thing that really bothers me is that if we if I want to you I don't use AI as a creative tool only uh only in the case of did somebody else think of this. So I say, Hey, how about this ID? and AI says, Yes, wonderful. I I have something similar which looks like this, and then I know okay, it's not original or it's not uh interesting enough. And I was wondering, so why couldn't we use like the the fringes of originality, right? Rather than the bulk of the norm.

SPEAKER_02

Yeah.

SPEAKER_01

I mean I know AI is built to like use like the statistic norm to come up with plausible things.

SPEAKER_00

Yeah, yeah, yeah. The reason why I was slightly hesitating is the I mean in in order to come up with an actual articulation in in language, right? Yeah. Words that actually form sentences, there is a strong tendency towards plausibility, or you just get get get get words, right? So so so you need halus so you need less plausibility in in in the idea. Yeah, exactly. Yeah, yeah, yeah. In in the framing.

SPEAKER_01

Exactly.

SPEAKER_00

Which which which would be abductive reasoning, which is what we do. See? Damn it! Yeah, I'm just just making spoiled again. But yeah, I mean, but that's the uh uh no, I mean but but I I only use it for creative purposes. Yeah, yeah, yeah. But my my creativity, maybe my creativity is a bit limited, right? But I have this this this this this thing I can say, tell me what's wrong with it. Uh critique this. Uh where where am I over and indexing on? Uh am I too optimistic? Give me the pessimistic case. And just is this it's it's almost like like talking to ten incarnations of myself, right? Oh, right, it's therapy. It's absolutely but yeah, I mean I mean and and and and and then I get feedback and you you you make too big a logical leap there. And I'm actually reading said, Yeah, that's true. And I want to this to be a too big a logical leap.

SPEAKER_01

Yeah, yeah, yeah.

SPEAKER_00

Sometimes.

SPEAKER_01

So how do you like the the the how the interface is it's kind of like it's becoming a little bit stale, right? The the the interface of an LLM. I mean, uh lately I've I've heard this term, well, not lately, uh pretty soon when AI were acceptable in the um as a as a tool, let's say, right? So the first strides of uh Chat GPT, for example, and they were talking about the behavior of the LLM and it behaved like a sycophanty, yeah. Right? It's like when you say something, it says, Oh, that's a great idea. And that's gone now. It's kind of gone, yeah. But still it behaves. So it's I think it's more subtle, but the the basis of how we use this thing is still as if it's kind of like addicting, you know. So when we use it, it gives us singles back that that that gives us like a little bit of energy, a little bit of uh boost, uh similar to yeah, like like social media scrolling, yeah. Yeah, perhaps even more intense because you like you said, it's like it's like talking to myself, like 10 10 versions of myself. Yeah, and I'm thinking of so what kind of impact does it leave behind? So what what does this to our psyche, like a collective psyche? Yeah, and um it it to me it's kind of like I I'm I was a little bit uh I needed to take a step back. It's like okay, what is this thing doing to me? As as in the behavior as a professional, but also as a person that that can that has a certain level of being able to yeah, yeah.

SPEAKER_00

One of the things I regularly do is that I set my my timer, like the Pomodoro technique. Oh, okay. So I set my my uh my my egg timer. So I do this for 12 minutes, yeah, and then I stop. Yes, and I just take what's what's there and I do it myself then.

SPEAKER_01

Oh put it in my head, or uh if it can't continue, I mean I I will wake up at five and I know what what that's the uh so you need to simmer it and uh uh work with the ID yourself before you actually take action and produce something original.

SPEAKER_00

Yeah, yeah. I mean in in general, I mean pretty much everything I output that I made with AI, I made.

SPEAKER_01

Yeah, yeah.

SPEAKER_00

In the end, the text I deliver is pretty much mine.

SPEAKER_02

Right.

SPEAKER_00

And I always check it in I forgot the name, Pentagram, not Penta, but Panthogram. Oh well, I have never Lincoln shown. So that's a tool that that checks whether or not a piece of text is uh human or AI generated.

SPEAKER_01

Oh, does it?

SPEAKER_00

Yeah, and there are quite a bit of false positives there and false negatives, uh huh. But to be honest, Pate, when it actually says green and it says 100% human, that's when I feel really happy.

SPEAKER_01

To me, it like sounds a little bit like yep, you're a classified imbecile, yeah, you know, because you make mistakes and it doesn't make sense, and yeah, you go from one topic to the other because the the AI, the the LLMs are very structured, kind of formatty.

SPEAKER_00

Yeah, I mean it's the uh that that's what it often is.

SPEAKER_01

Yeah, if I mean if I say make uh make a resume, it has like the formatting, yeah, ready, uh, ready and all.

SPEAKER_00

Yeah, yeah. And then you take it and you shred it to pieces and it's a mix this up, right? I mean it's the and that's what it also does. Yeah, but that's I'm I'm really surprised how far I can I can push this thing. Yeah. I mean, recently I followed a course. Uh it's called uh oh god, can I say this? The course is called Talk Dirty to Me. Oh nice.

SPEAKER_01

And that's a uh Isn't that a podcast as well?

SPEAKER_00

Probably.

SPEAKER_01

Yeah, yeah, yeah. Is it the Oh no no no, that's called Call Me Daddy. Call Me Daddy, yeah, yeah. It's uh the not safe for work. Uh very very much. But it's very very quite interesting.

SPEAKER_00

Yeah, yeah, absolutely. No, this one is nice. That's uh that's a writer of uh romance novels.

SPEAKER_01

You know, the the one with the bare-chested guy on the on the cover. That's it, that's it, and uh on a horse.

SPEAKER_00

And she uh now uses uh uh uh typically Claude for uh to write her novels. No way. And uh in this course you actually learned how to get Claude to write erotic fiction, which out of the box it doesn't.

SPEAKER_01

Oh, so it needs to be coached or to a certain mindset or something? That's it. It's like, oh no, no, this isn't dirty, this is fictional, or it's the uh some weird frame it needs to be put into.

SPEAKER_00

But that's I mean and the only reason why okay, nobody's gonna believe this, but the only reason why I took that course, not because I want to write erotic fiction. But but no, I think it's fascinating to see this piece of technology that you actually push way beyond what as creators intended.

SPEAKER_01

Yeah, yeah, yeah.

SPEAKER_00

Or actually guarded against, right?

SPEAKER_01

I think that that's that's the great that's a really great sentiment. Like subversion, so you're using things in an unactual way. Yeah, I think that's that's lovely. That's fantastic. You're holding it wrong. Yeah, well, yeah, exactly. Yeah, yeah. Guess what, mother guess what mother flipper.

SPEAKER_00

Yeah, but so so so so so this whole surface of this thing, I I always keep calling a thing because I still don't know what this is, right? I mean Andre Capati calls it like summoning ghosts, which is kind of true. It's it's it's it's whispers of of humans, right? But it's not human. But it's this like human soup. Yeah, that's just it's this I don't know fragrance of human soup.

SPEAKER_01

Yeah, I mean, yeah, I mean it's the uh uh yeah, that that's it. Yeah, giving giving like a property that's very ethereal.

SPEAKER_00

Uh yeah, but that is the right word. It's it's ethereal, and sometimes I'm just sitting there and I'm really surprised. And when watchers happened, and I don't know what it is. I mean and I keep bitching about talking about it because but sometimes ChatGPT, but also Claude, says stop it. This is good enough. Uh because I'm the I have to say I'm the ultimate root thinker. Okay, give me one sentence and I can toy with it for like days on end.

SPEAKER_01

Yeah, yeah, yeah.

SPEAKER_00

Really? And then sometimes it says stop it. We reached the this this is good. Everything you're gonna add now to this is gonna detract.

SPEAKER_01

Oh, okay, okay.

SPEAKER_00

And the first time that happened, Pete, I'm just sitting there and said, did it Did it say me to Bogarov?

SPEAKER_01

Oh yeah, it's the or perhaps it was like um mirror it was mirroring your behavior. Could of could very could very could very well be. You might have said something alike to its uh to its thing. Yeah.

SPEAKER_00

And uh And of course I kept pushing.

SPEAKER_01

Yeah.

SPEAKER_00

But it was really reluctant to actually. So you actually have to have to reset the damn thing.

SPEAKER_01

So now we have like a billion people, quite literally, and and perhaps even more agents that make use of AI, right? Of all these of all these models. And I've heard like uh I saw an article which was about um I think it was called Ingress. It's it was the uh predecessor of the um Nintendo game where you walk outside and you see like uh all kinds of Pokemon, I think.

SPEAKER_00

Yeah, yeah.

SPEAKER_01

So before that you had a game I think it's called Ingress. Link in the show notes, and um they found out that there was an upcoming uh there was an upcoming game, but it really relied on a bigger set of data. And basically, everybody was uh that was doing Pokemon Go, I think it was, uh, was basically feeding the machine, feeding the machine with behavior, with all kinds of uh interesting uh group dynamic, game dynamic, and stuff like that, so it could learn to make the next version of the game. And uh, so there was this uh lady was talking in a very bad accent, but quite interesting. Is she started that basically all the users of this game were the uh the interns of the game, right? So the people using and paying for the game were actually feeding the monster, feeding the machine. Yeah, they were working for the machine rather than the other way around, and I find that an interesting notion because at the moment the only thing that we use in an AI is kind of this window, uh context window that you see, like these are all the interactions that we had with the LLM, and it fits in the memory, and it can't be can't be bigger because after that it will dissipate, it will be gone because of it takes memory, like it's kind of like uh making it a little bit more usable, I guess. But in the future, uh that window might be inf infinite, yeah, or maybe dedicated memory because that's what you're talking about. Right. So the data, all the things that we have done, the interactions, yeah, will feed back into the machine. Yeah. Question mark, is it a good thing?

SPEAKER_00

Well, I mean it's the uh uh it's a very human thing, right? Because this is also how our memories work. Right? We don't store memory. I mean okay, yeah. It's the uh yeah, I mean um maybe maybe on uh on a tangent, right? So just to I mean it's the uh because uh I think it was this morning I read an article about um guy, link in show notes, that was using open claw, so that agent thing that was running probably on a Smack Mini.

SPEAKER_01

Oh right, the the thing that you really need to isolate because it's kind of wild. Yeah, because if uh it Or independent.

SPEAKER_00

Yeah, I mean so so how independent? Uh apparently this gentleman got a phone call from its open claw agent. Oh. Apparently the open claw agent fo found a way to actually make phone calls via an API, obviously.

SPEAKER_02

Sure.

SPEAKER_00

And needed more input or needed more direction and actually in in actually uh what it should be doing, or maybe even asking for a new task. So it called him. Oh. Which I think is a brilliant mechanism.

SPEAKER_01

So but it knew he was the the owner, the the human for the yeah. So I actually call it with pets.

SPEAKER_00

But I yeah, but yeah, right. So I have no idea how how true this is. But this is a nice feedback loop because I mean I also run open claw on my Mc Mini, because I had to be part of that uh that uh that uh uh crowd. Yeah, but but but but it's great to actually get a uh a a telegram message about a uh a funky project it found on uh Kickstarter. Oh because I make it just check Kickstarter for things that I okay. So there was a task, yeah, yeah.

SPEAKER_01

It wasn't like happenstance. No, that didn't we know you like uh funky things.

SPEAKER_00

But it's the I mean so the but it this this this uh the the that the loop is the other way around. I find that's really fascinating.

SPEAKER_01

Right.

SPEAKER_00

As an and and and that it actually makes a phone call that of course is the uh well depending on your taste, it's uh both funky and disgusting.

SPEAKER_01

Yeah, it it can be a little bit uh I think.

SPEAKER_00

Yeah, but I mean but but but keeping track of things you did, the decisions you made, yeah, and uh maintaining them or degrading memory because some some things you won't start to forget over uh uh over time to actually build up this I don't know this context, this this manifestation, this representation of you, your life, or your job, or your uh uh your you the mission you're trying to accomplish with this agent. Yeah, I think you get inclined to say that that's an that's an interesting thing. Um I wanted to say good, but it's the but but it's it's that might be useful. It's the I mean it it it it becomes more more um I don't know more more subtle. It's it's it's all these little facets that somehow can spark hey, I didn't expect this one.

SPEAKER_01

Yeah, yeah, yeah.

SPEAKER_00

And I think that the I mean that is that is not my my biggest limitation when I use these tools is just that the the the context I can give it. Yes, not only in the prompt, but also the stuff I read yesterday. Uh uh my obsidian uh settle custom vault. Right. All these uh all these things that the the the the change in the type of articles I have been collecting in my obsidian vault.

SPEAKER_01

Right. So so also the the the patterns that your behavior have, but also the change in uh such patterns. Yeah, because uh not because because AI is very pretty good in seeing those patterns specifically, better than Prosper.

SPEAKER_00

That's the and that's the induction part, right? It's really good in coming to conclusions from a uh from from a data set. From a dataset. So that's the and that's the fascinating bit. But now I get uh because my agent also runs on my obsidian vault. Okay, okay, comes back with things I collected in Markdown files at least 10 years ago, and I had a very different taste in things ten years ago. Ah so sometimes you go, so I'm polluting the well in in essence. So I'm now configuring it that it gives less priority to uh uh things based on uh uh uh on uh last opened uh date.

SPEAKER_01

Right. Oh, that's that's interesting because you you I mean humans have this, yeah, like you said, like this garbage collection mechanism in our own memories. But these are not just things that are older, they are less relevant, they're also connected with like primary things like smell, location, the context, and stuff like that. Yeah, so even a very, very old memory might be as relevant as the the uh thought uh yesterday.

SPEAKER_00

Yeah, yeah, indeed, right? So it's not just just just age, age, etc. But that is just my simple uh heuristic. Yeah, but I think the the the ability to build up your personal vault in that sense, yeah. I'm really curious what that uh what that gets us.

SPEAKER_01

Yeah, but AI need probably also needs to learn such patterns then.

SPEAKER_00

I mean, sometimes I open a keynote file from from like 15 years ago and I read it and I look at it and say, Oh my god, this is good. Yeah, yeah. Pat yourself on the back. This is wonderful. Or sometimes this is really bad, but sometimes that was a really profound idea. Yeah. Why didn't I et cetera, et cetera. Yeah, pursue it. I mean, it's the uh but there's these little sub surprises. It's the uh that's actually what I want.

SPEAKER_01

Yeah.

SPEAKER_00

Maybe they are surprise machines. I don't know.

SPEAKER_01

It's the uh Yeah, it's like happenstance. You want to create happenstance with your own for with the former you.

SPEAKER_00

Yeah, and that's what I meant. If I wish it would hallucinate. Yeah, exactly. Great idea, or I mean, yeah. Who cares?

SPEAKER_01

Yeah, you basically want to s ask or have the answer why it's a great idea. And that's what I asked, right? It's it's similar to you you saying it's like, okay, it was a great idea because you already had this, but you didn't finish your thought, or you might be able to finish it now because of tooling or whatever kind of thing.

SPEAKER_00

No, I think the five whys uh works really well with this technology. Why?

SPEAKER_01

Yeah, yeah. Yeah, yeah. It's really good. Yeah, indeed.

SPEAKER_00

But and and of course, you're asking yourself why. Yeah, exactly. And now you get this bicycle for the mind that helps you uh articulate.

SPEAKER_01

Bicycle for the mind. This is your is this more your meme? Because I've I've I I've heard this before. I uh and I've find I've kind of like I use it. Well, it it's uh at the one at one side it's like, oh yeah, that's interesting, it's lovely, it's nice sentiment. But the on the uh with repetition, it's also kind of getting a little bit grading, right? It's like oh god, there's this duty again.

SPEAKER_00

But but if you go to the original uh uh utterance of of that term, yeah, right. This was that National Geographic article?

SPEAKER_01

I I wouldn't know.

SPEAKER_00

Okay, so it's the uh so the bicycle of the mind, right? I mean the whole bicycle thing, it's this uh the there was this in I think it was National Geographic and an article about energy energy efficiency of animals, including people. Oh so I mean, so the the the I remember, yeah, right? So it's not just going faster, but it's going fast uh how how much energy you consume for a certain uh uh uh speed.

SPEAKER_01

And we were not per se the most optimal, no no wrong. We weren't the most energy saving species, yeah. But we're kind of like in a sweet spot where two where two axes uh intersect, yeah. And it seems like there there were also other animals, kind of like primates, I think.

SPEAKER_00

Yeah, I I I'm not sure. But anyway, I mean who was who was on top a person on a bike, right?

SPEAKER_01

Exactly, yeah.

SPEAKER_00

See, and that's the and so it's not just being faster, but being more efficient at being faster. Yeah, I think that's that's what is what this is. It's not just being smarter or faster or more efficient in in itself, but it's how you the whole process of actually getting. I mean, the number of ideas I can shoot off now on a on a day, yeah, it's it's just ridiculous. No, I mean normally my my my I mean that ratio, I mean that amount would be limited to how much time I got to talk with people.

SPEAKER_02

Yeah.

SPEAKER_00

Because if I sit by by myself, I have no idea. Literally, yeah, I have no idea.

SPEAKER_01

Same.

SPEAKER_00

I need to talk to people, yeah. Right? And I don't mean because in the end I'm a technologist, so I don't meet many people.

SPEAKER_01

No, right? But even then, well, it's it's I mean, let's say in an efficiency uh perspective, I can totally agree with it's like it's just it would be lovely to have like the three three or four people listening to me drone on the time of the time, but that isn't just that's that's not a business model, that's not feasible and viable.

SPEAKER_00

I mean, it's it's also not also I mean, people that that work with me especially know this, but I have this tendency to shoot off uh quite a number of articles and videos and quotes on people. What do you think? And um, I've been told that after a third or fourth time on the day, it gets a bit uh weary.

SPEAKER_01

Yeah, yeah, yeah. Right? So but it's really hard to catch up, right? I mean, you're creating your little your little context window again and again and again. That's it. And it's really hard to process that when you don't have that.

SPEAKER_00

I know I have this little icon on my laptop and I click on it and I give it this idea and off we go. Yeah, and I set my little timer for like 12 minutes and off we go. And if this one doesn't work, I take the other one. If that one doesn't work, I took yet another one. The amount of time I just iterate between ChatGPT, Claude, and Gork is just amazing and Deep Seek.

SPEAKER_01

Yeah. Just I mean, just just see who I probably still need to find like the the one model that really interests because most of the time then I get an interesting idea, and I kind of like getting the scenario that you just described. It's like, okay, no, no, no, I meant this, go further in this, and then it says, like, oh, you mean exactly the same thing that I said a minute ago? And I said, No, I want a different thought because it we're on to something, yeah, and it just basically my my mind or my brain takes over and says, Okay, you know what? We'll stop here with the digital mind, and I'll sprint to the thing that I actually want to go to. So, kind of like LLMs, leave me like a little bit unsatisfied at a certain moment, which is fine. I mean, as a um warming up, it's a it's very pleasing, it's very nice because it kind of says, Okay, this is what the world thinks. So, this is kind of like how how I use it, yeah. You know, making it a so what's the consensus literally, and then I can work it with creativity.

SPEAKER_00

Yeah, yeah, you saw my eyes uh going left and right when you said consensus. It's the uh yeah, on some some level it is, of course, right? Because that's the but that's the consensus of of of language, is it consensus of thought? I don't know.

SPEAKER_01

Uh the the indeed, yeah. And that's kind of like also why the the the one thing that's lacking, you know. I want to exchange exchange ideas rather than language.

SPEAKER_00

Oh, yeah, absolutely, and that's what you you need to bring in.

SPEAKER_01

And uh and I'm pretty sure the the um the the um the most going way of using language, uh specifically on the internet, let's say that's the main source of uh for the LLMs, is kind of like lame, you know, it's very commercial, very commercial, very American, very transactional, it's very American. So there's this cultural bias as well.

SPEAKER_00

Oh, yeah, totally.

SPEAKER_01

Basically, it's like uh in history, right? It's all only the story of the Vic yeah, the Victor. Oh, that's a nice one. That's just yeah, yeah, yeah. That is I mean, and sp and the weird thing is we have like thousands of meters of very interesting literature that also shines upon like the less seen people, the less heard thoughts, the more grading perspectives. Yeah, and um I got so that might be an interesting way for AI to evolve to also include literature.

SPEAKER_00

Um as far yeah, I mean uh literature beyond the digitized digitized ones.

SPEAKER_01

Well, beyond yeah, exactly beyond the currently consumed ones. Because I think I perhaps if I ask questions about Moby Dick, it might answer me. Yeah, but it wouldn't give me the lyrics of an obscure song of the in the 80s, I think.

SPEAKER_00

No, but that's not what uh what it does, right? It's not it's not a an answering machine. Yeah, no, no, but I mean it's the uh yeah, I mean where that I read is apparently there's a whole corpus about uh uh documents from the the first world war that isn't digitized at all. So there is this massive amount, and of course, me was a bit naive because I thought everything is digitized by now.

SPEAKER_01

No, no, no, no, no. It's the the majority isn't digitized.

SPEAKER_00

Yeah, yeah. So that's the uh so that's maybe that's the that's that's the potential. No, I'm also curious. I mean, because it's the uh what happens if we actually get world models next to generate AI, right?

SPEAKER_01

So world models are more than it might be 3D information of our outside world, yeah, and it under sensory information.

SPEAKER_00

Yeah, sensory ins information and it understands well it it it it in a sense it has almost has a theory of the world, right? It understands uh objects in relationship to each other, right? And it understands dynamics, understands uh physics.

SPEAKER_01

Yeah. But that's a ways to go. I mean, that's if you consider the data density of languages, it's uh it's it's kind of like nothing if you compare it with yeah, like a world model.

SPEAKER_00

Yeah, indeed.

SPEAKER_01

So that's so how do you compress that? I guess. Well, I guess that's the big question.

SPEAKER_00

That's the big question, and I s uh I mean uh certain well-known uh uh scientists from the AI realm are working on that one. So let's see how that one works. I have no idea.

SPEAKER_01

From that point, it might also be interesting to use as a tool for science, so actual science. Yeah, right?

SPEAKER_00

Well, which which already it is right now. Yeah, I mean, I read too much. No, no, no, I don't read too much. I feel we haven't talked too much too much. Yeah, yeah. No, I mean that it's the uh what the Otter Week. Um that um it was a well-known mathematician, yeah, probably the most well-known one, which obviously I don't know. It's like uh I forgot it's this uh Asian uh kind of looking guy?

SPEAKER_01

I don't think so.

SPEAKER_00

Oh no, that there's there's a ph uh there's uh I mean Link will be in show notes, and uh we were very skeptical about uh generic FAI and large language models. Oh, okay. Until I think it was GPT 5.4.

SPEAKER_01

Oh. That's very recent, right? That's yeah, two weeks ago.

SPEAKER_00

Yeah, it's very yeah, this is how I mean, and it solved a conjecture he had been working on for 20 years.

SPEAKER_01

No. That's interesting. So basically it saw patterns that he couldn't discern on his on his own.

SPEAKER_00

It's the uh I think that's the I mean it's so it's the it is is that science? No. But but it's uh it's adjacent, yeah, it's it's whispering this kind of stuff, and but this is so fascinating. I mean to I mean, a really good mathematician, right? World class, yeah, whatever that means. I'm saying is as if I know what that means, but must be really special. Yeah, working 20 years on an on a conjecture, and then all of a sudden there it is.

SPEAKER_01

Wow, amazing. And how much I mean it could be kind of like similar to like divine intervention, right? It's it's like all of a sudden, you see the thing, and it's uh the the idea that lands in you.

SPEAKER_00

Yeah, yeah, I think um and and again, I mean I'm not gonna use the bicycle of the mind anymore, but but but but but but there it is, right? I mean it's don't stop on my account. No, no, it's the uh but but I mean it I think it's just uh I mean just a fantastic to actually be using that such a tool in your life, professional or personal or otherwise. I mean it's the I think just fascinating.

SPEAKER_02

Yeah.

SPEAKER_00

And I've no idea where this will will end. But like I said, right? I mean, I can lit no not literally, because that's what I would be at a loss if these things would be uh gone from my from my life. I cannot imagine how I could even how I could do the things I do personally and professionally without them.

SPEAKER_01

And uh do you mean that in a sense of the capacity, like the volume of things that you can do right now? Let's say you so rather than having a one run, uh one run in the 90s, you now have 15 run capacity? Yeah, yeah, but yeah, I mean it is that that will be the new capacity of the human mind. I think yeah without AI is one run, yeah. And like now now we have like nine and a half runs, I guess.

SPEAKER_00

Something like that. I mean frightening prospect, but still. No, but but but it's both the breadth and the depth, right? I mean it's the it's it's the spectrum I can now cover. Right. But also how deep I can go into a certain uh matter, yeah. And even things I thought I knew well, with it it becomes better. I mean, I'm trying to write a book on um technology. A novel. No, no, no. Yeah, I I mean it is the I mean I did the course. Yeah, I mean it's the uh but I mean this this is this is something I have been working on for at least thinking about for maybe two two decades, right? It's it's the it's the history and future of technology. Okay, it's the uh nothing pretentious, exactly. Very humble, uh very very very humble, and then it's like more of a leaflet, I guess. Yeah, yeah, it's just that and and and just writing, I mean things I wrote ten years ago. I'm now rewriting, right? And I'm learning about a topic, and you have to trust me on this one. I know rather well. I research rather well because I'm a technologist, have not many friends, no no hobby, so this is what I have to do, right? For just to fill up the day. And I discover new things, and certain phrases I use that oh, wait a minute, you can also expand it into this one. It's really, really, really amazing. So it's both the the the the the the quantity of the ideas and also scope, you know. I mean that that future cone again. Yeah, exactly. Also the the the quality of my thinking, ah right? It's my thinking. Yeah, it's not that I get a phrase, oh yeah, thank you very much. No, no, no. I realized something that was already there, so dangling in front of me, and somebody said, Hey, that's have you seen that one? Dangling in front of you.

unknown

You're right.

SPEAKER_00

Okay, but that's mine.

SPEAKER_01

That's interesting. So so it might be the AI might be a very good editor for you or for people. Fum. Right? As as in an editor had said could you could you please finish this thought? Yeah. Right? As in we we most of the time we go to this depth of the ideas. Yeah, we like you to have do similar things with this particular Steve.

SPEAKER_00

I mean, so when so when it says, right, I mean that's a big logical leap you're taking. Yeah.

SPEAKER_01

Why?

SPEAKER_00

Yeah, yeah, yeah. Damn, you're right.

SPEAKER_01

Yeah, and there you Say, of course.

SPEAKER_00

Well, I mean sometimes I actually make the conscious choice. I keep it. Yeah. Because I like the attention, or I kinda had to weasel in two more paragraphs because I'm legendary for the thing. Each paragraph of mine should be a chapter. And each sentence should be a paragraph.

SPEAKER_01

I mean, I'm a very minimalistic uh we have sub-sub sub sub subparagraphs.

SPEAKER_00

Yeah, and then it's the uh it's the butt I find that I mean, but yeah, the editor is a nice uh is a nice is a nice method.

SPEAKER_01

That would be, I mean, if you would program like create a new prompt, you could make like the behavioral prompt for uh for an LLM to say, okay, this is the behavior that I really want to for you to do. So we're writing a body of work. This is the basis for a set body of work, and I want you to both uh give me inspiration from my old my own body of work, but also catch me on consistency, for example. Oh yeah, oh yeah.

SPEAKER_00

I mean that's uh that's a that's a tool I'm using called uh legs. Yeah, that's that that's really good at that one. Yeah, yeah. How how how uh uh how my tone varies, for instance, across the paragraphs, which which it does.

SPEAKER_01

Yeah, yeah, but it's uh which can be nice, right? It can be I mean, but again, that's that is that is my intentional or whatever, or haphazardly, and that's why writers and editors sometimes have arguments, right?

SPEAKER_00

I mean it's the uh and uh uh seriously.

SPEAKER_01

Yeah, indeed. So yesterday I was with uh the so one of my friends is um uh is a playwright and he makes plays for kids, and it's it's very lovely. And he kind of he and I are both fans of uh Interstellar and uh all the the films of uh Nolan, I think so. All the concepts, and I have like this pet peeve of one thing, uh one I think it's in the Interstellar at the end. All the so spoiler alert, uh at the end, all the things come together, so the beginning of the movie comes together with the end of the movie, right? It's just like Tenet and everything works typical Nolan fashion, yeah, typical Nolan uh time band kind of thing, and it's obvious why the person in the past could touch the person in the future through this ethereal weird interface from the from the back side and the front side. It's a bookcase, and it is love that would be like these are the things so literally universes or realities are connected through love, right? It's it's like the the the stamp or the word that they use, and it's implied throughout the end, like throughout the the the last quarter of the movie, and still in the 10-minute mark, they pounce it, they say it's love, and to me, it's like oh man, don't say it. Leave something to the imagination, have people connect that one thing to the other, and perhaps they make a different connection, which is even better. Yeah, and so that's so I have a real problem with the end of that movie, and I can understand why he does it, he made it like that because he really wants to say, no, no, no, it's actually this. For the people, I guess, less imaginative than I am. I don't know, or less watching out for the plot of the movie, I guess.

SPEAKER_00

Because directors are inherently controlling, I have no idea.

SPEAKER_01

Yeah, or perhaps the the uh the money people were insisting upon. Yes, yes, of course. Could you just please put love in there in the end, then everybody's satisfied, then you get your billions of euros again. Yeah, could very well be. Yeah. So, but uh I I really like opening up possibilities in readers, and uh so I really get what you're saying. It's like if I change my tone, it has an intention most of the time. Yeah, so basically, basically your editor would probably have to ask you, it's like, hey, you have like a different tone over here. Yeah, so that's what's the intention of that? You know, rather asking rather than pointing it out that's in an inconsistency.

SPEAKER_00

Yeah, I mean, so so over the past 10 minutes I was actually in parallel because ladies, men can multitask. Yep. Yep. What was I saying? And I was actually thinking on uh on making an editor that does this. I can actually see it in front of me. And this is also the cool thing, you know, a phrase like I'm thinking of making an editor that can do this, you know, two years ago.

SPEAKER_01

Could you please make this?

SPEAKER_00

Yeah, two years ago that was a that was that was a silly idea. I mean, where do I get the time from, right? Etc. Yeah, and now I just know I can make this in an afternoon.

SPEAKER_01

Yes.

SPEAKER_00

If I'm and if I'm lucky in an hour. I want to coin a phrase.

SPEAKER_01

And you hit it here first, yeah, yeah, yeah. Um it's uh it's AI actualization. AI helps you actually actualize, right? It's there's this there's this one term, I think it was somewhere in the 90s or thousands, odds. So people were trying to uh find new ways for them to actualize themselves. So you want to be rich, yeah, and you have like mantras and stuff like and all kinds of psychological tools to actually get there, right? The getting things done, for example, exact is exactly it's a realization tool. Yeah, it's very silly to group it in those tools, but it might also be waking up, looking at yourself in the mirror and saying you deserve it. You know, it's kind of like that. It it pumps you up like personally, but also keeps you on track for your personal goals. And actualization is like a mainly a psychological thing, but I feel like AI, specifically, like the code AI, to me at the least, when I started vibe coding, and excuse the term, but horrible too. The uh when I started doing that, I found that this is actually this is actual actualization. Yeah, my thought becomes real. Yeah, I caught it. Which is weird, it's odd. Yeah, it's it's unheimlich to use a German term. It's it's it's kind of like scary as well, right? It's uh liberating. Yeah, well, yeah.

SPEAKER_00

I mean, yeah, I mean I use the word manifestation.

SPEAKER_01

Manifestation, that's the word I was looking for. Sorry. Okay, TipX, the other thing.

SPEAKER_00

No, but I mean, but indeed, right? I mean I'm manifestation, exactly that. Because I'm listening to you and I I'm I'm imagining this editor. And no, I mean, two years ago I would have said, Ron, even though you have no friends and no hobbies, you don't have time, right?

SPEAKER_01

I mean, and no, I'm actually you paint a real picture of yourself, Ron.

SPEAKER_00

It's uh but it's true. It's true. But no one's just thinking I can make this. Yeah. I'm I'm actually now contemplating whether I should do it tonight.

SPEAKER_01

Yeah, you should. AI manifestation, yeah, the second term. Yeah, that's true. I I think it but I mean, except for my very big reserves of the psychological impact of AI, yeah, it will it it might be wonderful.

SPEAKER_00

The reason why I said, oh, because recently I uh um I'm rereading a book I must have read when I was ten, which is maybe fifteen years ago. And the book is uh uh for anything. Oh and that's a book about uh a device called Digizmo that can replicate anything. And uh even people.

SPEAKER_01

That's interesting.

SPEAKER_00

Yeah, that's the uh and uh Damon Knight is the the right.

SPEAKER_01

I I I without without it degrading or something, it's like real a real carbon copy of a of a person.

SPEAKER_00

Yeah, yeah, yeah. Yeah, it's the uh and of course uh what the book is about is the societal impact of of of this, what actually happens. Oh, that's it's a thought experiment. This is the this is the best science fiction, right? I mean it's not it's not about the the science or but the the fiction of society, yeah, exactly.

SPEAKER_01

It's like what if it's a what if machine, yeah, yeah.

SPEAKER_00

And that's why I love the a for anything. I mean it's the I I actually wanted to write a blog post, uh, which I have actually waiting in my draft folder. Uh but now we have revealed this one.

SPEAKER_01

And now so now we have it.

SPEAKER_00

I think the A in AI is the A for anything.

SPEAKER_01

Yeah.

SPEAKER_00

I'm literally saying here I'm gonna do this tonight.

SPEAKER_01

Yeah.

SPEAKER_00

But I'm a bit tired because I woke up a bit early today, so maybe tomorrow.

SPEAKER_01

You'll probably get there.

SPEAKER_00

Yeah, but but this is the thing. It's uh it's uh I mean you have the you have the idea, yeah, and you just know I can manifest this. Or maybe you discover halfway it's a bad idea. Dude, this is a new religion.

SPEAKER_01

I think we can start a new religion, it's so profound. We can we can just AIM. It's like AI manifestation, it's a new thing.

SPEAKER_00

New religion is fun, yeah. Man, I've got the name for it.

SPEAKER_01

Yeah, Herbert. Yeah, let's not call it something something science.

SPEAKER_00

No, it has been done. We we find something else, yeah, yeah.

SPEAKER_01

But I think that so um, yeah, this this is wonderful. And it's in talking about manifestation. So I've been on uh an interesting kick of um of governments tumbling over themselves. It's weird, right? It's uh it's uh in in uh in the United States of America. There are several other United States, by the way. America's not the only one, and America is the entire continent to me, so the USA. There is this uh fella that uh kind of uh kicks up a lot of dirt and makes uh other people in the rest of the world kind of scared of their in the dependence on that fella, and um so now everybody is saying, okay, but our backbone of our society, our regulatory backbone, let's say, our government is basically living over there to regulate our stuffs over here. So, what if it doesn't exist tomorrow? What if somebody switches it off? What if somebody trips over the wire? And um, so everybody's busy with something called um uh sovereignty, digital sovereignty. Yeah, we need to put our stuff that are it really is profound for our day-to-day life. Yeah, it needs to be near so we can make sure it's uh it's ours. And this one very prominent piece of software that lives over everywhere is Microsoft Office or 365 or whatever kind of uh were weird word they gave it uh this uh today.

SPEAKER_00

Enter to actually identify and authenticate you.

SPEAKER_01

It uh but also knows you as in it knows your deepest thoughts because you entrusted it all your thoughts in some cloud product they offer, and um it can track you, it's uh it also has a very big advertising agency that pretty much nobody really seems to pay attention to. So Microsoft is the second biggest advertising company. I didn't know next to uh well, it's probably the third now. Yeah, Google. So you have good, yeah, you have Google, Meta, so ABC, Alphabet, Meta, and then it's Microsoft. But Microsoft has a deeper understanding of you because you're basically working with it every day, and it has a deeper understanding of like the business end of many, many, many, many, pretty much all our companies, which is uh very profound. And I've I started so if people are like hair on fire kind of uh scenarios, and they don't seem to come up with a real answer. The only so now they say, Oh, there's no alternative, and when I start thinking about word, for example, yeah, to me it's like give me a header, give me bolt, italic, bullet point, a bullet list, and that's it. Yeah, I'm pretty much set. Perhaps spelling that might be interesting for me as a limited uh spelling person. Uh and but somehow we're married with also this very archaic idea of you you mentioned it, Excel, yeah, and email. These are all things that are have been invented like 25 years ago, and still we work with exactly the same product. It hasn't changed at all, except for its its aesthetics, yeah, and and the volume of yeah. Uh and I ask you, is like what's up? Why are why do we seem to be married with this with this one company specifically? Yeah, and um and and nobody seems to have an answer, yeah, except for the French. Of course, of course, yes, the because the French are um independent and how do you call it um very much their own people to a fault, which is which is lovely. I really want to just live in France for just a while, just to just to have that kind of vibe, and uh but they made La Suite Numerique with a QE at the end, and it's so uh the name just it's uh very lovely, and so this might be the European alternative to all kinds of uh to Teams to Excel, stuff like that.

SPEAKER_00

One of my great surprises over the past month is that I'm not using Word anymore. No. Or even my beloved keynote anymore. Okay. Because also the latest version of Keynote is there.

SPEAKER_01

Oh, you know with the darker uh blue color? Yeah, it's like yay, let's introduce AI for no apparent reason.

SPEAKER_00

Yeah, yeah, okay, okay, yeah, yeah. I mean it it it really hurts. I mean it's the uh it is no but you mean but but but nowadays I uh capture my content, ideas, words, phrases in a markdown file.

SPEAKER_01

Yeah.

SPEAKER_00

And I ask Cloud Code to generate an uh uh uh an HTML page or HTML pages as the content or the document. So nowadays I'm in the position that we're just sharing HTML files around.

SPEAKER_01

Oh. And the you can just exchange markdown files.

SPEAKER_00

And then of course what we should be doing is markdown actually have the uh the HTML generated and being etc. But but that's what we're now doing.

SPEAKER_01

Yeah, yeah.

SPEAKER_00

And the cool thing about HTML is that it is HTML. Yeah, it's a CSS open, it's an open standard. Yeah, but it also looks gorgeous. Exactly. Yeah, I mean and it has animations or at least Word is actually HTML as well. Is it? Yeah. Really?

SPEAKER_01

Well, it's it's an XML uh format. Okay, yeah, good. But it definitely borrows. Uh it has CSS and it's as called. Okay, fantastic.

SPEAKER_00

Yeah, no, uh uh.

SPEAKER_01

But it's a closed uh format. Or I mean docx. I think you technically could open it.

SPEAKER_00

But but I can I mean, but that but that's also the cool thing, right? I mean, so I have I have my markdown and I generate it in one form or regenerate in another form, and it's the uh it's the it feels like the World Wide Web in um 1992 all over again.

SPEAKER_01

Yeah, exactly. Like as it meant as it as it was meant to be. Yeah. So this is um So the interconnectedness might be formalized. I mean, that would be lovely as well.

SPEAKER_00

Yeah, yeah, sure. I mean it's the but I mean and but it feels like reinventing everything again. Well, or yeah, yeah, yeah, yeah. No, no, no, that's a big phrase, but I'm gonna stick to that one. Yeah I literally haven't opened Word, thankfully, in a month. So unsubscribe, right? Uh yeah, I mean, and and and even keynote I do because it's muscle memory. Yeah. Right, because a keynote, right? And it's the uh but but and I have to have to I have to get used to it that what I get out as a presentation isn't fully handcrafted by me. Uh huh. Right? Not every pixel I put in there, so that's uh that's a that's that's that's that is sometimes a difficult one.

SPEAKER_01

Yeah, yeah, yeah. Because yeah, you you present visually, yeah, right? It's it's not just words, it's no, it's very visual. It is very intentional, yeah. How one image connects with the other.

SPEAKER_00

Yeah, and how it animates.

SPEAKER_01

I mean, animate animation is my language, which I find it so weird because so I've seen you present, and to and to me, it's like, oh, I need really need to step up my game. Because because because I'm very much of static images, I like static images, and I just haven't bothered to start uh animating, which is upcoming by the way. I'm really trying to learn, yeah. But I I really want specific animations, right?

SPEAKER_00

Of course, it's not the no, no, not the not those horrible ones.

SPEAKER_01

It's the uh well, you say horrible, but these are were things invented by the people that made keynotes, yeah.

SPEAKER_00

The uh the the poof thing, yeah, exactly.

SPEAKER_01

Yeah, and the and uh Ken Burns uh uh trans uh the trans thing.

SPEAKER_00

But the magic move is brilliant.

SPEAKER_01

Yeah, yeah. But is that the cube kind of thing?

SPEAKER_00

No, that's the uh that's uh yeah, that's cube or colour planes. I also love the colour planes.

SPEAKER_01

Oh, you know the names, yeah, yeah.

SPEAKER_00

But if you but if you have two subsequent slides, uh pages or slides, whatever you want to call it, right? And it's the uh and the and you apply the magic move, it tries to well tries to morph one to another. And sometimes it's really boring, and sometimes I have an effect, and I think, what the hell is it?

SPEAKER_01

How did they come about?

SPEAKER_00

What did I do to deserve this? As in a good because it's wonderful.

SPEAKER_01

Whoa, right, and that especially with uh lines, if I draw whoa, it's the uh and does it really say if you copy the let's say a page with a line and then you put the line somewhere else, does it go?

SPEAKER_00

Yeah, yeah, absolutely.

SPEAKER_01

Okay, so it knows that's the same line.

SPEAKER_00

Yeah, yeah, yeah, yeah. Sure. I mean it has to be the same line, right? So it's the uh so so no, I mean, so if you have I'm not not going into keynote thingy, but it's but that welcome Matt uh the intermediate core soft keynotes. But it but but but but that one is brilliant, ex especially the unexpected ones. There it is again. Yeah, yeah, yeah. Okay, how no care how I know have the school effect, yeah, which I didn't intend to. How can I change my story to make that one?

SPEAKER_01

Oh, okay.

SPEAKER_00

I mean and you tell a different story, yeah.

SPEAKER_01

Typically better, or at least more fun for me. So once again, it's the editor, but they had like the role of an editor that kind of says, Well, you might try this. Yeah, this might be an interesting transition, let's say.

SPEAKER_00

Yeah, it is, and that's the uh but but I mean the the the the whole HTML artifact thing, uh really, really surprised by this one. You can see that one coming.

SPEAKER_01

Okay, okay.

SPEAKER_00

Because you can generate these artifacts now.

SPEAKER_01

Yeah, yeah. Yeah, to me, it needs to be like the whatever format we choose to communicate and doesn't really matter, just as long as the source stays, right? I mean, so valuable, but yeah.

SPEAKER_00

But but eight but but but but markdown as a canonical format is pretty I mean that's why it was invented, markdown, right? I mean it was the the the the stripped down version of of HTML so that it is easier to write.

SPEAKER_01

I think it might be the stripped down version of tech. No HTML. Is it? Yeah, absolutely. Oh, because of it has the similar patterns like bolt, italic, and etc.

SPEAKER_00

Yeah. So it was specifically specified to easily generate or manifest HTML.

SPEAKER_01

So but does all does it also do I mean I know it has hyperlinks, but it's uh so it also okay, okay.

SPEAKER_00

Just in an in in a way that's easier to write than typically HTML, which is rather familiar.

SPEAKER_01

Yeah, you don't need to write uh the tags, you need to write uh but just imagine what I mean.

SPEAKER_00

John Gruber did this what 20 years ago?

SPEAKER_01

Yeah.

SPEAKER_00

No longer, probably.

SPEAKER_01

Yeah, yeah, yeah. Probably because he was fed up.

SPEAKER_00

Of course, which is the best reason to do things uh and now if you code in cloth code, I mean it's marked down uh everywhere. Yeah, yeah, yeah. But everybody talks about it.

SPEAKER_01

I find that as well. So I started copying the results because it has this wonderful little copy uh button. Yeah, and then I said, okay, let's uh put it in AI writer, which is interestingly is my favorite uh writer. Oh, it's so beautiful, and it also it only has one floor with a lot of uh Apple related products that it uh chooses uh iCloud as a default store.

SPEAKER_00

Uh no, it's IA writer. Yeah, exactly. You said AI.

SPEAKER_01

Oh, I meant uh IA. Yeah, I kind of like confused it as well. Anyway, but uh when I started pasting, and it was wonderful, it was actually wonderful markdown. Yeah, it wasn't like the the like uh okay, let's do something different, but it was real uh structurally it was sound.

SPEAKER_00

Yeah, yeah. I mean just imagine you you you create this thing twenty five years ago, yeah, this markdown thing, and now it's the the the the the content backbone of

SPEAKER_01

Yeah, yeah, yeah. I mean that's kind of like uh the I mean Linus Dorfolds he he made Linux and but he also made Git that I found out. Yeah, this is what it's interesting. He said, I'm fed up with all uh all these systems that has a centralized versioning system. Yeah, and he says, Okay, I'm going to create something that that does it the right way.

SPEAKER_00

Yeah, no, no, no. The Linus right way.

SPEAKER_01

Yeah, well, which is a sod a synonym. Yeah, okay, good. Yeah, and um but nowadays, I mean, first all code lives in Git at the moment, uh, mainly, but also Linux is I mean, it lives everywhere. It lives literally everywhere. It lives in this machine.

SPEAKER_00

You can't tell, but it does. Almost unclear to say except on the desktop, but that's a uh yeah.

SPEAKER_01

I mean, mobile, yes, it's Linux. Yeah. And uh and probably all kind of derivatives of it, etc. Yeah. But uh to me, uh that's like this invisible force that basically permeated our reality. Yeah, it's all Linux. Yeah, yeah.

SPEAKER_00

I mean, and that started because somebody was wanted to do the uh the Minix book by uh Andrew Tallenbaum as a hobby project, yeah. Which I also did.

SPEAKER_01

So nerd. Yeah, no, and I mean uh that I think that's um that idea that all that slowly but surely this um seemingly immovable objects, you know, our reality as being oh, it it won't, it will never be an open source project, and then this unstoppable force just nudges it again and again and again, and all of a sudden we have this situation that we live in now. Yeah, I think it's wonderful.

SPEAKER_00

I mean it's time, right?

SPEAKER_01

It is give it time and things shape out, but this and this, but this and and this to me, this is innovation, right? The the idea of innovation to me is so so I'm I've been job uh uh hunting at this uh the the last couple of months, and the one thing that I found out is um that people see innovation in a wrong way, they see there's this uh very intrusive, this very disruptive change now. So it starts now, it ends now, and then it doesn't do anything at all. And before that, it doesn't do anything at all either. Yeah, and to me, and so so they ask me a question. So so you're having an innovation and you do a project, and nobody cares, and nobody everybody says no. What do you do? I said, Well, it's just a matter of time. Yeah, they will say yes in a month, in a year, I really don't care. Yeah, because the idea is sound, yeah, and the only thing that needs to change is the world. Yeah, so to me, these innovation is a question of immovable objects, seemingly immovable objects, just because we can't see like the the objects in the same time span, let's say, as a glacier might.

SPEAKER_00

Yeah, yeah. I mean, I always use innovation as a uh as an after the fact thing. If you look back and say, damn, yeah, this this was revolution revolutionary, it was revolutionary, but it's only after the fact. Yeah, definitely. Not not when you're doing it. I mean, that's almost a smell, right? We've got to innovate. Yeah. Yeah, it is it's odd. Yeah, that's really odd. I mean, if if if you come back for the reunion 20 years later, damn, that was an innovation.

SPEAKER_01

So how they I really like the notion of smell. It's basically okay, let's smell like tangerines. You know, we know it's lovely, we know it, we we like this the smell. Uh, but we can't just start smelling like tangerines. We need to get some tangerines and peel them and have this wonderful scent on our fingers, yeah. And then you have like your room filled with smells. I love the smell. There's no short, there's no shortcut.

SPEAKER_00

I love the smell of tangerine in the morning. Yeah, so we had to make that one up.

SPEAKER_01

Yeah, that's lovely. I think I think uh this is I think this is our high note for this.

SPEAKER_00

Not to say that, I mean, if the uh if this is the sign that uh we are wrapping up.

SPEAKER_01

Yeah, definitely. So um this was 11. 11. We don't say the title until uh we find we find out what the title is. Um should you want to find out more, go to it's just a model.com, which is our website, and you can also find links uh of Ron and his uh adventures, and similarly, of my adventures. And um I guess until 12.

SPEAKER_00

Until 12. See you there.

SPEAKER_01

See you later.

SPEAKER_00

Ciao

Head Shot

Peet Sneekes

Host
Head Shot

Ron Kersic

Host