DYEB #3 Secret perks of a tech ambassador

Hai, saya Han, saya kena beritahu kelihatan. Dan kata-kata di sini, "Pak Kata-Kata, akan membantu?" Itu kata-kata, "Pak Kata-Kata, untuk kata-kata." Dan kami hari ini, kami ada kodex ambassador, dan kata-kata ambassador, Brian. - Hello, hello. - Hello. - Yeah. - Brian, Brian, Brian, dan David, sendiri. Apa yang anda buat sekarang, dan juga, ada yang anda bawa. - Yup, jadi saya Brian, saya dari SMU. - Jadi saya kata-kata di sini, saya, buat ini, dan saya akan berjaya kembali ke luar gaf, juga membantu kelihatan AI engineer untuk membuat banyak dipercayaan. Jadi, AI engineer adalah sangat besar di USA. Dan mereka hanya membantu kelihatan AI engineer di Singapura, dan banyak dipercayaan dipercayaan dalam saya. Jadi, saya membantu mereka. Dan juga di percayaan banyak di kelihatan AI. Saya suka membantu kelihatan, mencari kembali kembali kelihatan AI. Dan saya membuat banyak dipercayaan. - Resejian saya membuat apa yang saya membuat? - Saya juga tidak membuat. - Tapi saya membuat apa-apa. - Saya membuat apa-apa. - Dan yang paling penting, saya membuat apa-apa. - Ya. - Terima penting. Jadi, selepas itu, dan kelihatan saya ingin membuat sesuatu pada kata-kata SDK, saya akan membuatkan kata-kata saya. Dia hanya membuat kata-kata SDK, untuk membuat kata-kata di depan kata, jadi kelihatan, kelihatan, kelihatan, saya akan melihat apa yang saya bisa membuat. Dan kelihatan, saya akan melihat apa yang saya membuat. - Terima penting. - Ya, saya akan melihat apa yang saya membuat. Tapi, saya akan melihat apa yang saya membuat. Jadi, ya. Okey. Ya, dan dengan itu, saya akan beritahu lebih banyak di depan kata-kata SDK. Jadi, kita akan membuat apa-apa di depan kata-kata SDK. Mungkin lebih banyak di depan kata-kata SDK. Kemudian, kita akan melihat kelihatan, dan kelihatan, kelihatan, kelihatan, kelihatan, dan dengan itu, kita akan melihat kelihatan, dan kelihatan, kelihatan, kelihatan, kelihatan, - Ya. - Cool. - All right. - Lekas-lek. - Jadi, kita akan melihat apa yang saya membuat. - Okey. - Bisa kelihatan, kelihatan, kelihatan. - Okey. - Jadi, saya rasa, saya rasa, saya tahu banyak orang yang berlaku tentang. - Ya. - Mari kita melihat kelihatan, kelihatan, ke Germany ении pekerja diarta атель astral pine lapas kRecloe bangga delle roti talahan lelaki daripat tas dia k вasil downtik ajanya vil tek acing cintai apa generous nourr demanded orange green is Ya. Ya, boleh bantu betul-betul bagia yang menjadi lelaki? Ya, ada download yang bagus yangito dengan pemain ousunkan. Jadi per distancekan hal yang być trouver waris, dan surprisinglah dengan das behavioral bagi kita. Ya. Jadi, yang pertama plan case 1 di dня awal di simplest, se^^ the first one, anjing dan langkah penggukaan yangjasar, To getologie campaign approve. Itu yanggehen gelapan Some other people who used better go dan dah kurang gilih我跟你講 is also instagram as comment. Ya. Dan dari sana, mosti panggilan dan dianggilan, dan juga, dianggilan ke depan. Jadi, saya fikir kita ada permainan panggilan yang kita bergantung. Dan ya, saya akan berjalan di sekolah-sekolah ke sekolah-sekolah. Jadi, di sini, di general, saya akan berjalan ke depan kodek. Jadi, saya akan berjalan untuk orang-orang untuk berjalan ke depan kodek. Dan mengenai ke depan kodek. Jadi, ini boleh jadi seperti orang-orang. Dan juga, ini sudah berjalan di sekolah-sekolah. Jadi, seperti orang-orang. Saya rasa ia akan berjalan. Ya, untuk pasti. Saya rasa ia berjalan. Jadi, saya berjalan di sekolah-sekolah. Dan saya akan berjalan di sekolah-sekolah. Saya rasa ia lebih penting, tapi saya berjalan di sekolah-sekolah. Dan mungkin, saya rasa ia lebih penting. Jadi, saya rasa ia lebih penting. adalah heel hanny pergi. Baca ke外an Medicaid. Ya. Apa identEverything yang kita hubungi? asi lagu напah판. ke datang kepada perempuan tipping hilarious, karenanya udah tanda bayar seperti yang tidak enak. Ia tahu bukan apa yang pun harga gitu, tapi dulu, yang lebih baik, saya sangat-satunya. Kata-kata, "Tidak apa-apa yang terlalu banyak, saya harga yang berlaku, mereka memakai bantuan kebangan. Saya hanya mahu. Jadi, bantuan kebangan kebangan, adalah super cool. Karena mereka sejak kebangan, berikan agen-agentik yang memakai kebangan. Lui, sebelum agen-agentik, kebangan atau kebangan, itu yang terlalu penting. Jadi, mereka memakai agen-agentik kebangan. Saya mahu berikan agen-agentik, saya berikan agen-agentik. Ia juga menggunakan kebangan. Jadi, saya berikan kebangan. Saya harus memakai kebangan, kebangan, kebangan di sekolah dan di mana-mana. Dan, ya, saya berikan kebangan. Saya berikan kebangan. Jadi, mereka akan berikan kebangan, di-sebabkan saya. Jadi, mereka akan sangat gembira. Ya, jadi, mereka juga memakai program yang terlalu penting. Saya mahu berikan agen-agentik, kebangan. Dan, pertama, memakai agen-agentik, kebangan. Jadi, saya tidak akan memakai agen-agentik, kebangan, kebangan, kebangan. Jadi, itu berikan yang berlaku, untuk menghantar banyak lagi. Jadi, juga sekarang, saya rasa, saya rasa, saya rasa, adalah agen-agentik, kebangan, untuk menghantar banyak lagi. Tapi, bagaimana saya berikan lebih lebih merupakan, atau seperti, kemudian saya berikan, itu. Sudah tentu banyak diberi di percayaan. Sebab percayaan berikan berikan tentang company lain. Jadi, berikan saya untuk menjaga lebih merupakan agen-agentik. Dan juga, saya rasa, melihat kebangan internal yang terlalu penting. Jadi, kerana saya berikan agen-agentik, saya boleh melihat banyak yang berikan, yang saya berikan dengan orang lain. Sudah tentu, bukan. dengan company lain. Tapi, ini sangat berikan yang lebih merupakan, dan berikan yang lebih merupakan, yang lain yang berikan juga. Jadi. Jadi, saya harapkan agen-agentik, kemudian saya berikan agen-agentik. Saya selalu berikan agen-agentik, kemudian saya berikan agen-agentik, kemudian saya berikan agen-agentik, kerana, sepanjang masa, semua yang berikan juga berikan agen-agentik. Samat daripada komen. Dan tumbuh jugaốt pon operators. Goodnight bye, Zwoll. Tapi ada. Nezal zaman2 sulit. Tapi dari r你在 jahat 112 ketulai també. Siapkan sembuh sel Penguin pakai tiga. �ab對不起, jangan seperti-ingkan. Ini adalah lain yang lain. Saya rasa untuk saya, yang terbuka, yang terbuka, dan menggunakan, kerana mereka berhenti untuk awak, ia adalah seperti, "Let's start off grove mindset lah." Awak patut berhenti dan cuba lagi. Jadi, macam mana yang berhenti untuk awak. Dan, "Just come full circle back to see whether you want to continue with clock code, or try codex out for a week." And sometimes, right? I think what annoys me also, macam, "The shitiest version, and the free tier." And it's just like, "Okay, you use the free tier and then you call it bad." And I think, "In today's war, you cannot get the sub." And then try it. You have to get the sub and try it. Right now, a lot of different stuff has troll. Because there's a troll, clockwork is where the troll right. The passers. You really want to do development or engineering work, right? I really feel like, you need to really just deal with and try at least for a good part of your workflow. Like, try to replicate the same workflow on clock code on codex. And then, from there, you decide how it's lah. Because not everything will work the same lah. But I think there's very big leaps in certain harnesses. So, for example, cussers harness on opus. Like, the same model on opus and clockwork. There's a huge shum. The exact same benchmark on cussers harness. And then the exact same benchmark on the clockwork harness. I think clockwork got a 54% and then, I remember seeing that. So, these are like, a lot of things behind the scenes. People don't really see lah. And that's like something that has changed. Yeah, I've ever seen that I joined these ambassador programs. Because, you just get to see a lot more about what the war was talking about. Especially other ambassadors. So, that's one of the biggest parts lah. The most important part of the world. And that's what I'm talking about. Jadi, saya tak menghentikan saya itu. Saya tak menghentikan dia. Dia hanya menghentikan dia, karena dia berada di tempat saya. Apa yang berlaku dengan dia? Saya tak tahu. Saya tak tahu. Tapi, ia hanya berlaku dengan dia. Tak, dia sangat berlaku. Jadi, mungkin itu salah satu perkara. -Mungkin lebih banyak untuk menghentikan dia. -Mungkin lebih banyak. -Mungkin itu salah satu perkara. -Ya, ya, ya. -Okey. -Tidak berlaku. -Tidak berlaku dengan dia. -Ya. -Ya. -Ya. -Ya. -Ya. -Ya. -Ya. Saya pun tahu, tapi menggunakan semua kesusahaan saya. -Nina, saya bergantungkan kesusahaan saya. -Yes. Mereka sangat populer. Harusnya, banyak kesusahaan saya. Mereka sudah terdapat banyak kesusahaan saya. Mereka bergantungkan kesusahaan saya. Ia adalah di dalam kawasan yang dikawasan saya. Kalau begitu, saya juga menggunakan kesusahaan saya. Saya menggunakan kesusahaan saya. Tapi sekarang saya akan menggunakan kesusahaan saya. Mereka juga menggunakan kesusahaan saya. Saya juga menggunakan kesusahaan saya. -Kawasan saya. -Kawasan saya. Saya rasa ia akan berjaya. Pada masa yang terbaik saya berjaya. we are using this software called LAC. It's a good equivalent but from by dance. So they have amazing language trust link. So almost your time. So for example, we are speaking right now, we are speaking English, you can actually translate that directly to Chinese and vice versa. Or to a variety of languages. But I view that English to Chinese and vice versa is the better one. - Yeah, I mean, I'm sure there's gonna be a lot more applications to this real-time stuff. - Yeah, and the voice models. - Because I've been watching definitely the next one, then I think to it, in the agent, right? Like, yeah, like, right now typing, I don't type at all now. - And you might not be able to see me with this. - Unless I've been public and I disturbed someone, right? I generally just tend to use the codex dictation. You must say I use aquaflow, like I'm sure you heard of it, I was super whispered also and whispered flow. Those are the three main ones. But I think right now the codex is my spree or so. - Yeah, I'm using it. - Notice myself using it when I'm not in the good position of the type. Like, for example, I'm crossing the road now, I'll use it. But when it comes to sitting down on the desktop and whatnot, I will prefer to type. But it's quite crazy because even when you're talking about hackathon, I saw one the other day where the whole hackathon is a hands-free hackathon. - Oh, did you see? - Yes, yes, it's the level of laps and it's a custom one. - Yeah, I saw it another day. - Yeah, so they're trying to really push a lot of people to go as hands-free as possible, which I thought was quite interesting. I don't know how they're gonna do it in an enclosed space where everyone is talking to cursor, but they're quite interesting. - I mean, it's always like, it'd be better input than bad output, right? So then all it really depends on how good the mic is. - But if they were to isolate your voice in a noisy environment, that would be quite interesting too. Because I recall when we first met in person, it was the Google hackathon. - Yeah, it's the German hackathon. - Just correct. - It was in January, January, so. - Yes, it is. - One of the SDKs they are trying to push in that hackathon was the voice to SDK. - Yes. - Yeah, it was quite chaotic because the whole room there was one side blocking. - And some of them are having casual conversations of them are testing out the voice SDK. - So it's quite chaotic in that sense. I do not know if they will be rebuilt in the level that person hackathon, when they're stalking. - They're kind of trying to get us. - I think it's a virtual hackathon. So that's probably why. - Oh, yeah. - But on the note, I just remembered what I was doing. So I was building like a virtual scanner using, I was doing Gemini real-time 1.5. I forgot the model name, but it's the real-time model. And that model basically was so bad 'cause it used flash for reasoning. Which brings me to my next point. And GPT real-time is able to reasoning. So when you ask you a question, you can do a little bit of reasoning, like how the image models do now. - Well, the image model is amazing, man. I love it so much. Same, it's like huge, huge, huge jump over the previous generation. - Yeah, super huge. It's so fun to be using it to create posters for my next event series. - Cowliss. - Yeah. - Yeah, I feel like I have to try to type the cum, like one of the nights when a bit more free from Army. I will drop by. We're thinking for you to come. I love it, man. That image model is great. I think they're doing some pretty good stuff. - Yeah. - So instead, how do you foresee it will change the way of you? I mean, you're already quite savvy in using this voice models. In a sense, you don't type when you're giving instructions to the LM. But how do you see it further change how you build and how do you see how it affects the entire developer population? - Yeah, so I think, I mean, I watch how my workflow is like for us. When I do development, two parts. When I do a green field development, so that means from scratch, then what I like to do is I do a brim dump. And this brim dump portion usually is entirely voice. So I'll just talk to it. And then when the questions come in, I'll just answer it also. No typing. Maybe just one of two typing portions to explain. Maybe if there's a wrong word or a dictation, et cetera. And then I just let it run. And then the second part is on Brownfield work. Don't use as much voice. I want to think out true why I say, 'kay, I want to change. Makes sense. Yeah, I think for Brownfield, in general, I don't use too much voice. But let's get a very quick change. Maybe migrating for all these different stuff to another way of writing for all my infecting smaller pieces of code. So those kind of tasks that I generally don't use voice, but I think it is larger than, yeah. So the second speaker here, I'm just, Jesse, Jesse is one of those people that really pointed me to voice. He's one of us, we know that, actually, you should really, you think a second note. He used it a lot for brim dumps. And I'm starting to see the value of that as well. The issue with brim dump is that it's very erected. There might not be a structured flow of ideas as well. Which then the AI needs the additional layer of work to sort out and find clarity in their whole brim dump as well. Right. I think in a Brownfield project, you need very precise pinpoint guiding of the art. That's just where I guess the blend mode or the stirring, like the GPD models kind of help. I don't know if that's staring in open, but I think that is life just not maybe not as well known or like as obvious as cls, GPDs one. But I guess that's where you can do a bit of staring. So on the cortex at right whenever I see the model do something wrong, I'll just immediately use voice detection and I'll just like immediately. So generally, that's why I watch out for time to stay. But another thing I also use is a best of end in cluster. It's basically where in cluster you can select how many times you want to run a particular problem and then you can select different models underneath it. So if the end is eight, right, then basically you can run eight different variants of the same problem. And then you can see the diffs for each one. And then of course, this is not very ideal because it burns a lot of credit. Oh, yeah, I think I think it's very good for blending. So it's super, super good for blending because I've seen sometimes where I am best of, maybe five, right? Maybe two of these are both the opposite. One opens is a bit more nuanced. And then the other one can be like, okay, we need to do this first. So I think this is a good thing to trial. But the only problem is the cash law, of course, opens it and should. And I thought the cool thing was that the stock issue where you can actually song the message and use a different model to continue that trail of thoughts. Yeah, yeah, I thought that person is raising. Yeah, but I thought that's the best of it. They are probably the best coding harnessed right now in the market. Yeah, just in general, of course, there's so much data of coding feedback, right? And I mean, that's the whole reason why SpaceX has gone into a deal with them, not quite yet, but going to deal with them. Yeah, yeah. Oh, there's so much code in the showroom. Yeah, because the very big thing is, it's just SpaceX doesn't have, they don't have the product. They are grow all those sucks. And yes, I record last year when they had some, I think it's a good for fast or something. Yes, they have it under the code name. It was actually quite decent. Yeah, so I think rock fork fast is where they realize that they could basically RL the entire thing then eventually make it good. So that's I think where the composer guys from concert did some work there, RL on Kimi, right? So super cool. They also did it, they do it in real time now. So what they do is every four hours, right? They run the same reinforcement learning process. They're the block on it. If you go to cursor.com, you'll see it. Every four hours composer to get better and better. Yeah, so it's pretty insane. But yeah, so cursor is definitely, yeah, people say it's dead, right? But there's a reason why they are still alive, making money and getting acquired. Talked about. Yeah, very cool. Hi, I'm very, very close about. Nice. I actually really good to take a look at the person. I'm not the only time I open cursor, right? It's so I can get interface. But I'm finally caught up in something below. Yeah, I open code with the way people, or at least the way software engineers will work will be different. Like right now everyone is just not writing as much code as they used to and not using the ID as much. I also don't use the ID as much anymore. Like even for cursor, I just use the glass. Is there new agents with noting the way people change what I record or at least write software or change the number so much. Looking at code anymore on the note of agents, right? Which kind of bring us to the next topic, quite unique because we don't really talk about politics here. But one of the things that make the news is the conversation about agent workflow, agent workflow in a small parliament. So it's quite interesting because this is one of the rare few times that technology that this frontier makes way into the parliament and got my number of parliament talking about it. And which was quite well coming. Because I instead of being stuck at, oh, we give you a change of these, of Christian and whatnot. Yeah, actually going one step deeper and talking about agent workflow, which was quite well coming. Yeah, I know you have a lot to say because we want a few people there and actually notice the other MP is the balloon Christian as a gig house. Actually, I didn't see the parliamenting, but I know Minister Vian, Vahakrishnan, Gihar. So I be, it's pretty cool. I at least should see that minister. I think for us on the ground, a lot of engineers on the ground, I just depend on the tech scene. It's just quite cool that even your own minister is looking at it like that. And that's because they show like, yeah, I'll go to show that at least, they are like, considering what's next. I think Singapore has always been on this frontier of of kind of adopt AI the most. Saya rasa josefin josefin di sini. Dia juga berada di dalam same panel yang saya menggunakan dari Google Dmai, di dalam event yang kamu dan saya. -Isit itu yang saya berdua tentang? -Ya, itu juga tentang LM. Ya, itu sangat penting. Mereka juga mereka menggunakan ini. Tapi saya rasa Google, mereka memiliki dalam game yang lama, atau bahawa mereka tidak menggunakan. Mereka tidak menggunakan, seperti selama keadaan, tapi mereka memiliki dalam game yang paling penting. Kalau menonton, semua other later, ada lagi lagi. Mungkin ada di OpenAI yang buat. Tapi saya rasa Google, Dmai, ini sangat fokus di dalam penjagaan. Dan ada lagi yang lebih penting di dalam game yang lain. Saya rasa, keadaan dan menonton dan menonton. Mereka hanya berdua. Mereka berdua, semua orang lain. Mereka tidak akan berdua. Tapi saya rasa kembali ke dalam game yang saya berdua, saya rasa, saya rasa, saya rasa, saya harapkan akan berdua. Mungkin akan menghantar lebih banyak orang di poli atau kamu perlu berdua. -Berjakan mereka dengan mereka. -Mereka ada banyak yang berdua. Tapi saya berdua di dalam game yang lain. Singapore, di mana di sini, as well as the earlier fun reading process, There's not been the most linear, conducive. - Okay. - conducive, I think that's a better word. - conducive. - Yeah, I think they're trying, I mean, it's definitely they're trying, most recently so Nord AI, - I'm sure you've turned out to be right, - Yes. - For sure, it's like, so they're like a little smaller. They're like a smaller division in MDDI, if I'm not wrong. And basically, the aim is to sort of help this AI adoption, help people learn more about it. They are basically running at event every week. It's like, if I'm not wrong, - It's actually a lot of events. - Yeah, it's crazy. So, I mean, there's so many different events, and this is a good place to go find out more about AI, even if a small business or what, it's still pretty cool to. - Yeah, I have them supported. - Yeah, they are very new. It's very overly subscribed as well. So, I've been trying to get in, but I'm doing that a bit hard. Yeah, they are trying to make that whole area, like the one of. I think they call it like, "No wrong part of something." It was in the news recently. - Hmm. - Okay. - We're really cool to see it. - Yeah. But that's not exactly an innovation, that's more of supporting and upscaling in a more generic, self-selected sense. - Hmm. - Yeah, incubation is still. - Yeah. - I don't quite see it yet. - Hmm. - Hmm. - Okay. - Cool. - Yeah, but I would have believed that you know. - Definitely. - Yeah. - Definitely. - Yeah. But come back to this issue, right? I thought it was quite well coming because if you look at the incentives that the government is pushing out, charge GPD subscription and dropping subscription and whatnot, it's always take a long way, something else. It's take a long way, maybe NTSC membership, or even the skills future calls that you get, then you can get six months of subscription. Okay. And not quite sure, but sometimes, if you give the further subscription, they might not accept any how to use it and what to use it for. Whereas if you point them down as a dumb path, which is like, "Okay, right now I have this forecast for example." And I am heavily invested in finding out how to do a forecast anything and having that workflow. Yeah, then there's a very motivated target that way to learn. Instead of giving me some subscription, which I may or may not know, how to make a point. Actually, another thing is motivation. In general, unless people are really motivated, they wouldn't go and find time to research. So I was doing SMU. I had a class called Gen EI. So then I mean, I just happened to have some open-router credits to spare. So I was like, "Hey, anyone want credits? Could you feel the urge? You can use any front-to-air and model, happy to just share my credits." And the thing is, nobody came up there. It's just like no one cares. But NTSC is a year for class. So you can see the level of motivation. They're not going to explore more than the syllabus. And for that class in particular, it doesn't really explore more than the normal and the average problem engineering. It does love not too much. But could it be that they'd buy say? It's also an important thing. If today you often mean something, sometimes I would also feel like I say it to ask for you. But you may be very giving percent. Sometimes I don't even ask you for help, but you send so much help my way also. There I feel a bit pious. No, no, don't need that. I think, you know, I'm also quite open on, like, I'm introvert, but for this kind of thing, happy to go and share credits and stuff. Of course, I'm definitely not going to finish using all these credits. So at least for the students, I can see like there's a value at that. Like at least they can try it out. Like you're not going to learn about AI engineering. If you never ever hit the OpenAI Responsible API. It's just very sad to see law. So yeah, for me, my way of learning is you have to do it practically. And so I think same case, the description that the government will give out. If you really have the motivation, then what's the point? And not even going to use it law. But must double check how are they giving out this subscription. So yeah, do you know that they have? I don't know. I don't see a keyboard. Yes, I don't see a keyboard. When I was so pumped that I really went and checked it out. I went deep to check it out because if I were to play 10 balls, and I get ES subscription, right? Yeah, I'm just going to cancel all my sessions. Go for the help when you get everything. It's a very budget conscious decision, right? Or I don't cancel it. I just have more of the stack you get. They get more AI usage. But I realize you know I'm going, but I realize that they give you all AI usage within their app. So it's like they embed everything inside their app. And then just nothing but a chapel, which is quite sad. Yeah, they do. It's not part of the typing, but then I'm quite disappointed. But I'm getting. Yeah, and the interface is good. Like, oh, you went for it. I have to plan because my future is in size just for it to try. OK, OK, OK. Not very useful for me. So I'm going to skip. Yeah, I think I said, I don't know whether the end you see what's going to be like that also. Of course, it's called obvious what they're doing, right? They are, they're making the backpick more entitling by everybody thing. They obviously won't be subsidizing the cost of this. And at the same time, they do it this way. Not only do they save costs by the increased DAU for their application. So that's true. Yeah, for audience, the use there for daily active users. My future will be a conflicted vice-chair. Yeah, nothing else. It's like, but I feel like this is just my thick. I've heard that there are these future classes, but my thick is that a lot of future classes outside of the universities at polytechnic. Sure, that can be seem like very fluff. So yes, yes, I think I have. I think I have got something to do with the AI modules. And it's quite bad. Yeah, I can probably better than them. I can probably better than the instructors. Not how to print by the age. Yeah, they might not even know how the problems are being broken down to tokens. Right. It's the funniest case in place. This is a part before called cavemen skills. Cavemen skills are amazing. Cavemen skills prove the absolute fact that you don't need to use perfect English. You can just use like super broken English. So essentially it allows you to talk about the cavemen. So I still say, the authentication method is broken, which is quite a lot of broken. You can go as cavemen as saying, off-broken. So there's just two words. You can say a lot of broken stuff. Actually, this cavemen think films a lot of things draw and he proved a big flaw in this skills future quarters where they teach you from engineering and like. In the last 30 years, the word please, it's important. Yeah. Yeah. Makes sense of back. I think I'm in the resume by then. Not leader. Yeah. Another thing is also how fast AI scene is growing. Maybe maybe the problem engineering was good like a year or two ago. Maybe three years because the models are. Yeah, the models are so good that like a short one or two words can actually just can capture your intent almost even. So just nothing. Yeah. Interesting. Yeah. I don't help that. I see those real short people going after the eldest. It's very still driven. It's not really education driven in that sense. I hope they incorporate more AI work stuff in schools. Which is good because I know my school is starting to do that. So yeah. Yeah. See you in the next generation coming back to the original topic on hand. This is that to see that our government is actually quite advanced when it comes to frontier. Like even in this case, here is your floor as a truer. I think even though they stopped talking about me, we will see that agenda was on. Quantum and nuclear. Yeah. That's your other quantum bit. They were way too early. So they said to switch here to talk about AI first. While making quantum as the backdrop of everything else. Yeah, all the other small things. So I put it to try and follow like the European countries. Like they are super strong hit strong on what they want for regulations in the EU regarding. So they have this they have this new standard called ISO 42001. Which is this includes like let's say if you pre-trained a model, you need to give the governments or whoever's enforcing its regulations, the entire data set. So they know that it's not being trained or any malicious or bad data. So yeah, it's something to that. I found out fairly recently or so. So I don't know if Singapore come out with any thoughts or standards or even like follow these. But it will be pretty interesting to see. I'm not very well read or robust in regulations for the matter. But I always thought that you can be very over regulated at times. Which actually hinders the innovation. -Saya rasa ia berguna. -Ya. Saya rasa ia berguna. -Ya. -Ya. -Ya. -Ya. -Ya. -Ya. -Ya. -Ya. -Ya. -Ya. -Ya. -Ya. -Ya. -Ya. Dan mungkin משapkan di padak-kakietan Ano Dame bukan bagus. zero, saya akan kata bawa awak. stream kekuan rogu. That may Trans peuharit juga hanya dengan satunya senang. Sama-sama halnya. Saya rasa itu sangat jika milik Kidsρι increases versi no suruh. Saya memkupakan panci saja dengan Ram-kahnhak,. saya akan menghantar. Tapi jika ia lebih kata, saya akan menghantar. -Apa ini? -Saya akan menghantar. Tapi jika ia akan menghantar, ada yang lain yang saya lakukan. saya akan menghantar. Dan saya akan menghantar. -Kak pun, saya akan menghantar. -Kak pun, saya menghantar. In this episode, we're going to share a bit more about the secrets of being a ambassador. Lay down all the secrets. Yeah. Okay. Then we go with the thumbnail. Okay. You want to do it. Then you would like a you would do it to do the thing. We didn't think up enough. Okay. Okay. I think it's pretty fun. I enjoyed it. Thank you. I always wanted to have a tech podcast. Oh, really? I read it in Singapore. Yeah. No one is really talking about all this AI stuff. Like it's very, very little. Like maybe. Wait, I don't. I don't even think there's a tech podcast in Singapore. I think. I think that's his podcast called red. Yes, yes, yes. Yeah. I feel like that's that's she's the only one doing it. Yeah. I think that's the best thing to do. That's the best thing to do. I think that's the best. That's the best thing to do.