Algorithms + Data Structures = Programs - Episode 244: High on AI (Part 1)

Episode Date: July 25, 2025

In this episode, Conor and Bryce chat about AI, how it's changing the way we work and more.Link to Episode 244 on WebsiteDiscuss this episode, leave a comment, or ask a question (on GitHub)Social...sADSP: The Podcast: TwitterConor Hoekstra: Twitter | BlueSky | MastodonBryce Adelstein Lelbach: TwitterShow NotesDate Generated: 2025-07-01Date Released: 2025-07-25AI Poll ResultsAll of Conor's Vibe Coded ProjectsCursorClaude 4Vittorio's CamomillaGPU ModeADSP Episode 238: Recommended Podcast Discussions on AI & LLMsADSP Episode 239: Claude-Poisoned Dev Sipping Rocket FuelCoRecursive Episode 113: When AI Codes, What’s Left for me?My AI Skeptic Friends Are All Nuts - Thomas PtacekThePrimeTime - How WE Use AI In Software DevelopmentIntro Song InfoMiss You by Sarah Jansen https://soundcloud.com/sarahjansenmusicCreative Commons — Attribution 3.0 Unported — CC BY 3.0Free Download / Stream: http://bit.ly/l-miss-youMusic promoted by Audio Library https://youtu.be/iYYxnasvfx8

Transcript
Discussion (0)
Starting point is 00:00:00 How would you describe LLM coding tools like Cursor, Windsurf, GitHub Copilot, Cloud Code, Codex, and others having changed your dev workflow? A. It's rocket fuel. B. It's very helpful. C. It's mostly unhelpful. And D. Don't use them slash avoid them. And this is much closer to the poll that I actually cared about because for me it's rocket fuel. It's completely changed how I work. I asked on Twitter, Mastodon, LinkedIn, and YouTube. Welcome to ADSP the Podcast, Episode 244, recorded on July 1st, 2025. My name is Connor and today with my co-host, Bryce, we chat about AI, specifically AI assisted coding
Starting point is 00:00:46 tools like Cursor, Windsurf, and more. So now this is doubly a good transition because I mentioned the language app that either it's on my list of things to build, but I'm also happy for someone else to build it. And then also within that episode talking about the Kuda toolbox talk, and you've seen the demo for this, we're going to share our screen, folks, and we will make sure to walk you through this and there will be links in the show note for all of this stuff. We're going to talk about for the next hour plus, might cut it up into two episodes. We might make it all one episode, folks.
Starting point is 00:01:24 We don't know. I now have a website called it's just a subdomain off of my github.io and it is vibing projects. It's matrix themed folks and everything here is vibe coded. We I assume including the website including the website. It looks beautiful. And so the first one here is a summary of the polls. I did two polls. We're going to talk about that in a sec. But I mean, Bryce,
Starting point is 00:01:52 you've seen this and it's just it makes me so happy. It's I hope the future of like the landing page for like the library ecosystem. Basically, I might even if I have the energy, which I never do, I always say that I say if I have the time and effort I'll go back and I'll find a clip, but I'm pretty sure in that episode when we were talking about this Effectively a slide from your presentation. I was like this is amazing we need to do something more with this like I pointed out that like the Rapids logo wasn't purple and You know it was hiding the kootDF and QML and friends. And so at that time I had this vision,
Starting point is 00:02:27 I'm not even sure how, or specifically I articulated it, but there's a chance that I basically described this and now it's here and definitely, because it wasn't in your presentation, you were the one that mentioned the categories. And so now it's all very interactive, it looks fantastic. It took me, I think. It's beautiful. A few hours.
Starting point is 00:02:46 I mean, initially the one shot was pretty good, but it didn't look anything really close to this other than the fact it was a four by six grid. If we leave this and we come back here, we've also got a summary of the two polls. I know you saw at least one of them. I did, yeah. And the first one that I asked,
Starting point is 00:03:02 which I kind of regretted, not regretted asking, but it wasn't really the question I wanted to ask was what category best describes you about AI? Is it a maximalist? AI will transform everything rapidly and profoundly be pragmatist AI is useful, but needs careful deployment. socially harmful or D, Doomer AI poses catastrophic or existential risks. I kind of feel bad about the wording because I got AI to generate this for me. And in hindsight, you know, I think B talks about like being careful and C talks about social harms. And anyways, it wasn't the best poll, but here are the results across Twitter, Mastodon and LinkedIn. Twitter got 156 votes, Mastodon got 57, and LinkedIn got 147. Overall, if you total up the votes across the websites,
Starting point is 00:03:51 Maximalist for A got 7.7%, so that was the second least common. The most common was B, Pragmatist, AI is useful but needs careful deployment. That had 48.7%, so almost half of folks. Skeptic was C at 38.3% and Doomer was D at 5.2%. I got one comment, can't remember on which platform it was, but they were like, oh, you've biased the polls
Starting point is 00:04:16 by using the word Doomer, it's charged language. Listen, okay, I didn't come up with the term Doomer. That's the term of art. P-Doom is a thing, the probability that you think, you know, the Singularity is going to happen, we're all going to get wiped out. It's the term of art. So don't hate the player, hate the game. Now, getting to what you said earlier about the poll, the one problem with the poll is that it sort of conflates two things, which is one, how widely, like, I think what you're trying to ask about was like, how widely and rapidly will AI be deployed and useful?
Starting point is 00:04:46 Versus in this there's a second thing which the poll is sort of conflating into that question, which is the risk of it Yes, yes, which I so I wasn't a huge having read it in hindsight I can which is why I ended up sending a second poll because on top of it not being the best wording I honestly wish I just had a left maximalist pragmatist skeptic doomer and then let people infer what they will about it. That probably would have been a better pool. How did you vote? I voted maximalist. I'm so high on AI. It's absurd. This website we're looking at, I basically just gave it the data from the polls and I was like, I want a static page
Starting point is 00:05:25 and it unnecessarily gave me these pie charts and donut charts, but it kind of looks cool. I was like, give me some nice little animations and stuff. If you click on these, they hover. It's all not really necessary, but the ability to build static pages that summarize information, do I have it open right now? I built this yesterday in like 10 minutes. We have two different wedding lists, one from this Wish Joy site that people RSVP to, and then another one from our, what do you call it? Like travel agent that did all the bookings for people that wanted to book through her.
Starting point is 00:05:57 However, like those lists are not the same. Some people use their formal or like official legal names with the travel agent, whereas they go buy something else colloquially, which is what they used for the RSVPing. And so my fiance, Ashima, she asked me to like consolidate these and I was like, well, you know, let's have some fun.
Starting point is 00:06:13 So I just went and asked, I was like, can you build me a wedding guest like consolidator? I have two different lists. Some of them have exact name matches. So I want like you to auto match them, which is this auto match button. But then after that, I want to be able to like manually just like click on one from the left, one from the right, and then have those paired and then like remove them from both of those lists.
Starting point is 00:06:32 And so the one shot gave me basically what this looked like with different styling. The second question I had to ask was like, oh, like, please remove them, which I think it actually was already doing, but I just I didn't notice. And then also like the the wish joy RSVS.P. list. It include everyone, including the people that declined the invitation. So I was like, I need like a filter button and then sure enough, like it built in like
Starting point is 00:06:53 exporting to a CSV with it like 10 minutes of work to do this. And Shima is she's doing hospital work. So she's out of town right now. I like Google. What do you call it? Google meted. I was like, sweetie, you got to look at this. It's so cool. And it's like your ability right now. I like Google, what do you call it? Google meted.
Starting point is 00:07:05 I was like, sweetie, you got to look at this. It's so cool. And it's like your ability to build these tools, like probably it would have been only five or 10 minutes more work if I had just like skipped this step, gone to an Excel spreadsheet because I'm good enough with Excel. I could have done this, but this was like, it was, it saved me probably like five or 10 minutes. And it was just so much like the power that you have.
Starting point is 00:07:26 What would it it would have been five to ten minutes to do the data now so it wouldn't have been five to ten minutes to make a UI for it. Oh yeah yeah just to like get the all we need is like a unique list of names of the people coming at the end of the day. Anyway so just your ability to build these kind of like single page things is just phenomenal. Can you make sure we RSVP'd right? Yes yes you were on the list of folks that like so at the end of it too like there's a bunch of of like single page things is just phenomenal. Can you make sure we RSVP'd right? Yes, yes. You were on the list of folks that like, so at the end of it too, like there's a bunch of people that RSVP'd yes, but didn't book through the travel agent. So I think there was a list of 24 folks.
Starting point is 00:07:55 And so then I have a button that says add people that self booked. And then in the CSV, it mentions in parentheses when you export it that they were self booked. And so then we're going to just send that back to our travel agent because And also the wedding list people or the the wedding resort folks because they need this full list But just like you know like this kind of tool how often is this kind of tool would it be useful? In when you're just doing stuff and like after when I showed a shima this I was like I was I was just so excited I was like choose something like well what do you want to do? And then we ended up, if I take this at a full screen mode, she has to do like billing
Starting point is 00:08:31 codes depending on the type of work she does. And she practices in a bunch of different specialties. So she, she's a licensed family medicine doctor, but her specialty is public health and preventive medicine. But she also specializes in addiction. So she was like, it's a very complicated world. Can you just like get it to go search up all the OHIP, which is the Ontario? I actually don't know what the hip stands for, but it's like the Ontario Health Card.
Starting point is 00:08:54 I guess Americans aren't familiar, but like different countries have free health care and the doctors, they they bill stuff. And so at first it came up with all the specialties. But then she was like, there's so many codes, like, just give it the four that I need. And then it has some fuzzy searching stuff. It's not perfect. And at one point, she also was like, oh, it's like it's making up codes and whatever. And I was like, well, yeah, if you're asking it to go and search online, like it'd be better if we just had like a PDF that it could scrape.
Starting point is 00:09:21 She's like, oh, like, let me send you the PDF. Then we drop the PDF down. She said it's still not perfect and she needs to review it. But like already she when she was looking for like a smoking code and she was like, oh, it's interesting because in the PDF it's listed under community medicine. It technically, I think she said was a hospital or family thing, but they labeled it as public health.
Starting point is 00:09:39 So it wasn't perfect. But she's like, but this is the code and that is the price. And she's like, you know, if I spent an hour with this polishing it, like this will make my life so much simpler going forward. And she and like literally like her excitement for the first time ever. Like truly, she's like, I now understand why sometimes I have a hard time like getting you to come to sleep because like she's like, I can now think of like 20 different applications that like I could build
Starting point is 00:10:03 something like this that would make my job and my life like so much easier. It's interesting because I also have infected Ramona with the AI bug and we only have one like family chat GPT subscription. So we're using like one account. So like if I go to chat GPT, I can see like all of her queries too. And like every day there's a new thing that she's been researching. She just got a new tennis racket and it hits a little bit different
Starting point is 00:10:29 than her old tennis racket. And so she's been using AI to come up with a training plan that she could use. And she was like, I want some, it's a PDF that I can print out. It's got a lesson plan. And I didn't even have to like encourage her that much. Like I showed it to her, I showed her like how
Starting point is 00:10:49 to start using AI tools like a couple months ago and now she uses it for like a surprising number of things. Is she using just the chat GPT interface? Because I didn't know it could generate PDFs or you're saying that you hooked her up with? No, no, no. Yeah, chat GPT can can generate PDFs can generate almost any file type like you ask it for like an Excel file for PDF for a whole bunch of things yeah it can do a lot of that I think I think it does I mean I would guess that at least in some cases the way it's gonna do the PDF generation is because it has it has its Python scripting environment so it may just it may just like run Python could
Starting point is 00:11:24 generate the PDF. But I think it has a native tool for PDF generation. But for some other file formats and for some scripting tasks, you'll see if you're using the reasoning mode and you look at the reasoning trace, you'll see that it's calling out to a tool to do the work. Right, right. Yeah, I mean, Shima is so excited.
Starting point is 00:11:41 She basically was like, I need access to it, because I'm using Cursor in Cloud 4. And she's like, I need so excited. She basically was like, I need access to because I'm using cursor in cloud four. Yeah. And she's like, I need I need this now because she when we were building this, like I showed her that all I do is talk in English and I like this is HTML, CSS. I'm not I'm not a web dev. You know, I know enough if I need to fix something because the model can't figure it out. But like these days, it never if it messes up some alignment, you just say, oh, hey, sorry, the pie chart when you expand it is slightly outside the border some alignment you just say oh hey sorry the the pie
Starting point is 00:12:05 chart when you expand it is slightly outside the border could you just like fix that you don't have to even say like I know that what you want to say is increase the padding so that it doesn't do that but like you can just use English and and I really think I was telling her that like there's this layer of stuff that you need to know like you need to know command line you need to know get if you want to host something on some website and then you need to know like at one point she wanted a certain color and then she's like, oh, it's this color. And I was like, well, can you give me the hex code for that? Like,
Starting point is 00:12:30 you know, it'll it'll it'll we're not at the point where you can just drop a screenshot in a folder and then it can like, you know, figure out the color from that. Well, I mean, I tried it. I tried it. So like, it were at some point we will get there. But like currently at this point in time, like like, there's such a thin layer of stuff that you need to know in terms of, like, you know, if you're trying to host some website, you just, like, put an index.html at the root of some GitHub repo. I could teach her that stuff in, like, an hour. And, like, all she needs to know is, like, when a model tells her that it can't do something, ask them to write a Python script for you.
Starting point is 00:13:05 Like if it says, oh, I can't go and like scrape this stuff, it's like, okay, well just give me a Python script that can because there's no limitation on what you can do with Python and Python can do anything. And so it's just like, you know, anyways, to get back to the poll. So it's good that our partners are effective. I have a question about the diagnostic code.
Starting point is 00:13:21 Look up thing, can you go back to that one? The diagnostic code. Yeah, yeah. Sorry, I just. The OHIP billing one. Look up thing, can you go back to that one? The diagnostic code? Yeah, yeah. Sorry, I just wanted to talk about my... The OHIP billing one. Sorry, I can't hear what you're saying. Do you mean the billing stuff? Yeah, yeah, yeah.
Starting point is 00:13:30 So, okay, so my question was gonna be what is your data source here? The data source is that there was some PDF that you fed it in and then it just embeds the information about the PDF into this. This is just like a static page. Yeah, so all of these are just static HTML pages. Initially, we just said,
Starting point is 00:13:51 can you go get the OHIP codes for all the different specialties? And then we gave it one example of like a, I don't know, 957 or something like that. But then very quickly, Shima was like, well, it's missing a bunch, that one doesn't exist. So like, I don't know what's going on here. And then I mentioned the PDF thing Shima was like, well, it's missing a bunch. That one doesn't exist. So I don't know what's going on here. And then I mentioned the PDF thing.
Starting point is 00:14:08 And she's like, oh, and then she dropped it into WhatsApp, and I downloaded it. And then she was like, OK, this is way better. So yeah, it was a PDF. I actually thought it was going to build a Python script to scrape it. But apparently, it knew about some tool called PDF2Text on my Ubuntu machine that it just called directly and it
Starting point is 00:14:25 was like a I don't even think it was optical character recognition I think it was literally just like it just took the raw text out of the PDF file and so then from there it did something with that stuff. I wonder whether like I wonder how much more challenging this would be if if you wanted to like integrate it with a live data source. I mean, I don't know. I guess I would do that.
Starting point is 00:14:48 I mean, these codes don't change. Okay, they don't change that frequently, yeah. Yeah. I mean, I'm assuming. But it would fix the accuracy problem. If there was some online place that lists all these codes, but not in a searchable way, then you could just tell it, just get the data from here. Just have some JavaScript thing that goes and fetches this
Starting point is 00:15:15 and imports it. I imagine in the antiquated world of medical record systems, they do not have some API endpoint that you can go ahead and get a JSON object with all this stuff. I'm guessing that, and I think that's why this was the first thing that came to mind for her. She's like, well, one of the trickier things
Starting point is 00:15:32 is like most people, they only operate in one specialty. Like they specialize as a family doctor and that's all they practice. But because she, her specialty was public health and preventive medicine, the first two years of your five year residency is in family medicine. So right off the bat, you basically become a family doctor.
Starting point is 00:15:51 Then she did public health, but then she did a specialization. So she's operating. And then she also practices as a hospitalist doing remote hospital work. That's what she's doing right now. So she operates across four specialties. And she's like, sometimes I don't work in the hospital work as frequently as I work in addiction work. So like, I'm not as familiar with the codes. And like,
Starting point is 00:16:09 you really need to know your codes in order to build properly. It's like one of the most important parts. And so like, she does all the billing correctly, but it's just like a pain. And I'm assuming the fact that she handed me the PDF is like, that's what she's she's looking up in like this large PDF that lists every single specialty and all these different codes when really she knows that there's some smoking billing code or some, what was one of the other ones, like counseling. So you just type in C-O-U-N and all the different individual counseling
Starting point is 00:16:37 or things that mention counseling in the description, they just immediately show up. She's like, this is what I need. I just need something that can filter out across all the different codes and specialties that I care about what the billing codes are. And it's like, you know, and that's you hear, which is kind of what we're we're burying the lead of this.
Starting point is 00:16:53 You know, both of you and I are very high on this technology. Yeah. But when we go to the AI poll, so the second poll that I asked, which was actually closer to what I cared about a poll for devs. How would you describe LLM coding tools like Cursor, Windsurf, GitHub Copilot, Cloud Code, Codex, and others having changed your dev workflow? A, it's rocket fuel. B, it's very helpful. C, it's mostly unhelpful.
Starting point is 00:17:14 And D, don't use them slash avoid them. And this is much closer to the poll that I actually cared about because I, for me, it's rocket fuel. It's completely changed, you how I work and if you look at the overall and this one I actually so I asked on Twitter Macedon LinkedIn and YouTube this only lasted 24 hours So it wasn't as long as the other one 55 votes on Twitter 9 on Macedon 78 on LinkedIn and 120 on YouTube the number one answer Averaged was don't use them, avoid them at 35.7%.
Starting point is 00:17:46 It's pretty even with BNC, which is unhelpful and helpful at 24.4% and 32.5%. But in last place by a long shot is 7.4%. Kind of interesting. I know Mastodon is a smaller set size. With the other poll too, where the Mastodon results were very different than the other results. Very, very different. It's 78% don't use avoid them. Like LinkedIn and YouTube are also different,
Starting point is 00:18:10 but the thing I noticed about the Mastodon results both here and the other poll was that the Mastodon results have a very extreme skew to one perspective. Yes, so Mastodon seems to be anti-AI the most for whatever your definition of that is. But anyway, so this is the backdrop of like I released two shorter episodes pointing people to other AI content, podcasts, articles.
Starting point is 00:18:34 One of the articles was all my AI skeptic friends are nuts, which I thought was fantastic. It's where I've gotten the terms sipping rocket fuel and he also refers to Claude poison devs. I don't necessarily like the word poison because it's got a negative connotation But I it sounds better than like Claude supercharged or Claude enlightened But I definitely am like if it's if the word is Claude poisoned. I am super Claude poisoned And I I yeah, I want to talk to you and I want to talk to other folks Like I just saw on my podcast feed what happens when all the dev jobs are gone or AI takes all our jobs.
Starting point is 00:19:09 And it was from the co recursive podcast. And there's a ton of discourse around like whether these things are good. They're bad. They're useful. Not useful. And anyways that's what I just want to chat like how there was another podcast by the primogen and friends guys. There's like four of them TJ Dax and Casey something and Casey he said he was using them 0% and then the other ones were anywhere from like 40% to 80% and
Starting point is 00:19:37 Anyways, I'm just like I'm I am so flabbergasted on how people can find no utility. If your argument is that it's bad for the planet and I disagree with the legal implications of copyright and whatnot, OK, fair enough. But the utility, I think it's more life-changing than the internet. It's completely changed the way I work. It's going to be changing the way that all these professions
Starting point is 00:20:03 are working, in my opinion. Anyways, I'm curious to get your thoughts I think probably the listeners have heard enough from me. What are your thoughts? What have you seen amongst like the people that you're Working with on a day-to-day basis because you interact with more folks than I do like on on research. So yeah over to you Yeah, so I mean Let's first talk about the is it is AI come in for everyone's jobs Let's first talk about the, is AI coming for everyone's jobs? You know, I'm not sure if we've used this analogy before in the podcast, I don't think so.
Starting point is 00:20:30 But in the 1980s, the first spreadsheet software came out like VisiCalc and Lotus 123. I think those were the first, or at least they were the first to be commercially viable. And the accounting industry was in a panic. They were like, oh, there's going to be no more accountants. 20, 30 years from now, there will be no accountants
Starting point is 00:20:58 because the spreadsheet will kill accounting. And I mean, that's obviously not true. All that's happened is that accounting has become less tedious. And, you know, it used to be like before before you had spreadsheet software, and before you had like accounting software, accountants would just have books, they would have paper, everything would be on paper, you go back and you look at these pictures, where accountants would create these huge,
Starting point is 00:21:26 huge like physical sheets. You'd have a huge table where you'd have a physical sheet where you'd do all the accounting. You'd have cells, you know, rows and columns just like you would in a spreadsheet. And that's how you would do all the accounting work. And it was like awful. It was incredibly tedious and like a lot more mistake prone. And then spreadsheet software came out. And it didn't kill the accounting industry.
Starting point is 00:21:56 It just changed it. I mean, there are still plenty of accountants. I don't know whether there's maybe fewer accountants now than there used to be. But it's not like accountants now than there used to be, but it's not like accountants are going away. It's not like it's an industry that's dying. The same for actuaries, right?
Starting point is 00:22:12 Like the invention of spreadsheets and of other computer software that actuaries now all use today did not make the field go away, just changed it. It made workers more efficient and I honestly I think it made quality of life a lot better because I think if you were an accountant in like the 70s and 80s and you were leaning over this table like filling in you know stuff on this paper I think you'd be a lot happier to like be doing it in software. So I don't think that software engineering is going to go away. I think that the field's going to change. Every time, I think, in the history of economics
Starting point is 00:22:56 that we've seen great increases in efficiency, we've seen industries change, but not like, there's a couple exceptions like you know agriculture you know there used to be like a large percentage of the workforce was you know worked on a farm or I think in like I read an article the other day like in the 1880s I think it was that like 30% of the United States GDP was the production sale of firewood. That's obviously no longer the case. But I think generally every time we've seen great increases in productivity and efficiency, we've seen quality of life improve for people. And I don't think we're at a point where
Starting point is 00:23:45 And I don't think like we're at a point where like we're gonna live in some Utopian society where we don't need to have humans involved in like work at all So yeah, I don't think that I don't think that all of our jobs are going away anytime soon They may change though in their nature. I do wonder if there's an analogy though in your accountant story or a recap of history in that I do think that there are a lot of industries including software where there are kind of levels of work. You know there's the senior engineer and there's the junior engineer and specifically if you're a junior engineer that's building these you know spas single page applications and like it's just HTML and CSS
Starting point is 00:24:26 and the other ones I think about the first one that comes to mind is like there's lawyers and then there's paralegals and the paralegals are doing this kind of and so I wonder in the accounting industry back in the day was there like you know the CPA accountants and then like some other job that would like do some kind of manual whatever and I do think that like a lot of those jobs are going to vanish because you don't need to hire a web dev person if you're just for the fancy JavaScript D3 animated, beautiful, whatever. I'm sure Chad GPT can help you, but it's not
Starting point is 00:24:57 the stuff that is the bread and butter that they can knock out one of these in 10 seconds. So I do think that there is some impact and I wonder if similar things have happened in the past but it's just like, it kind of gets forgotten in the annals of history. And cause I don't know, I guarantee you that there's a bunch of listeners being like
Starting point is 00:25:15 it's not exactly just that like a bunch of accountants were employed and they changed the way they worked. Cause now it's like, there's a bunch of software engineers that are employed that are now like 10 times, a 100 times more productive with these tools and they don't need to hire junior devs and so now like a small team can get done a lot more with a lot less like human resources. So yeah I don't know what your response to that is. I do think that you know change means that some number of jobs that exist today will not continue to exist.
Starting point is 00:25:46 But I think that the software industry will continue to exist and thrive. And I think maybe there will be fewer jobs, but I don't think it's going to be a decrease in order of magnitude. And so there's a principle in economics, I can't remember the name, but basically the key of the principle is that when the cost of something drops significantly, you tend to see great increase in usage. This came up when there was all the scare about the DeepSeq model a couple months ago where they're like, oh, DeepSeq, they trained a new, cheaper model.
Starting point is 00:26:28 And does that mean that the whole AI industry is going to go bust? And the answer is no. If the cost of training and deploying and using AI models goes down by 100x, that's going to mean that there's actually going to be more usage of AI because it will be more affordable. In the case of like a web dev, I think if you want to have a website built today, if
Starting point is 00:26:53 you're like a company, you want to have a website built today, like a simple website, I think it's like 5 to 10k. And there's a lot of businesses out there that still don't have great websites. In particular, within the US, it's pretty common for companies to have a website. You go and travel overseas, a lot of countries do not have websites for things. This always bothers Ramona. She loves playing tennis. Every time we travel, she wants to go find a local tennis club, they never have websites.
Starting point is 00:27:26 Like, we're in the middle of nowhere in Austria or we're in Turkey or we're in Bulgaria and she can find on Google Maps like a tennis club and maybe they have a phone number but they don't have a website that has the details about how to contact them, et cetera, et cetera. And so if the cost of getting a website developed by a web dev drops from like 5 to 10k down to like $100 to $500, because now the web dev, instead of them having to spend like a week or two doing this thing, they can just feed it into an LOM. They can pop out like 10 times or 100 times more websites a year, and the maintenance
Starting point is 00:28:09 burden also decreases for them. Then they can drop their prices. And then I think that means that a lot more companies, a lot more small businesses, a lot more organizations will have websites. A lot more things will become digitized. And I think if you look at the world and usage of technology in the world, there's still like a very long tail of the world that's not digitized. And, you know, processes that aren't automated, that aren't, you know, handled in software, tracked in software, companies that don't have websites, etc.
Starting point is 00:28:45 There's still a really long tail of that. The total market for tech stuff is much larger than the currently realized market. So I think that if you're a web dev, you probably should start using AI and probably the change that's gonna happen is that your costs are gonna go down and that's gonna mean that you're gonna need to find, you're gonna have to figure out how to bring in more customers and your business
Starting point is 00:29:26 is going to be more of a volume business. You need to find a way to automate more of your business to be able to survive in this new world. Do you think this is the key reason behind the polls and kind of the don't avoid or use or you don't find it that helpful is that there's a bunch of folks that are worried about the impact It's gonna have on like our industry Yeah, I Think so, but I also I mean
Starting point is 00:29:54 one one of one of the like, you know Bryce's laws is that like inertia is king and in the software industry you know the the almost always the reason that we in the software industry. Almost always the reason that we make decisions in software is around inertia. Usually like when we're making like poor decisions. There's a lot of people talking about in the AI space that like distribution is king, right?
Starting point is 00:30:21 The distribution is the key thing. Like if you have distribution,, that's your moat. And I think that that sort of gets to the point here that if you're able, if you've got a captive audience, if you've got users already, even if you're just an AI, even if you're just wrapping some frontier models, the hard part is not developing. Developing the models is hard but
Starting point is 00:30:47 There's always going to be a new you know Great model, you know coming out, you know, six months from now, right? That's going to disrupt the the field and The thing that really gives you long-term survivability is if you've got a customer base that's already bought into using your thing. Because for them to switch away from using your thing, the other thing that has to be 10x better. The OpenAI doesn't have to have the best models. Because we use chat GPT as a verb, right?
Starting point is 00:31:26 Like, oh, if you're gonna use a large language model for something, somebody says like, oh, I'm gonna go chat GPT this. Because it was the first on the scene, that's gotta be by far the most popular place where people go to interact with a model, with a large language model. And so they've got a huge moat because of that. They don't have to a large language model. And so like they've got a huge moat because of that.
Starting point is 00:31:47 Like they don't have to have the best model. And I think that, you know, for many, many years I just used like them and VI. Like I wouldn't use an IDE. Like I learned my ways of doing things. I didn't want to change, right? like I was happy with how I was doing things Change would require work would require me to stop what I'm doing and spend some time to set up a new thing And so I was resistant to the change just because like I was doing something that was working
Starting point is 00:32:18 I didn't want to go and you know change change how I was working. I think that's the way for a lot of engineers So I think it's I think it's the way for a lot of engineers. So I think it's, I'm sure some of it is people being afraid of it. And some part of it, I think, is people not trusting it. People thinking, oh, it's not going to be good enough. That's a piece of it, too. It would be interesting to know from those people who
Starting point is 00:32:42 said that they don't use AI how many people have tried it. How many people have tried it and Then gotten frustrated and decide to not continue because it is certainly a frustrating experience. I've had frustrating experiences using cursor Both on the like the model doing dumb thing is I've had like times where like cursor has been buggy You know all sorts of things like that. And you know, it took me a couple days to like learn how to use it effectively. And there were certainly times where I sort of got into like a rabbit,
Starting point is 00:33:16 like went down a rabbit hole of like, oh, I'm asking it to do this thing. It's just not doing it correctly. And instead of me just going to do it manually, I'm just like I'm gonna keep hammering on this thing and yelling at this model until it does what I want because I'm very frustrated and like it should be able to do what I want. And so I wonder if some part of it is people have tried and run into some of the limitations and then decided like oh well it has limitations and then decided like, oh, well, it has limitations. And so because I was already skeptical, I'm just going to assume it's not good enough.
Starting point is 00:33:52 And that may be part of it. But I think a large part of it is just inertia. It's just people haven't tried it because it's hard to get people to try new things. Yeah, I will say, I mean, I think we've all been there if you're using these tools, is it definitely you can end up in these rabbit holes? And I guess, I mean, one of the thoughts that when you were saying that is that
Starting point is 00:34:18 I wonder if a part of my maximalism on these tools is that I am an expert in C++ and it's already such a painful experience like half the time. And like I'm an expert but just sometimes you run into some like if you're doing bleeding it cutting edge stuff with like template metaprogramming or constexpr metaprogramming it's just like the errors are awful and like it is literally like there are talks by like Vittorio Romeo and friends of like tools that help you diagnose and like you know he's I know Vittorio has written like a tool I can't remember what it's called link in the show
Starting point is 00:34:57 notes but like it will it'll like collapse the nestedness of your like template metaprogramming compile time errors so that like you can parse them better. And every C++ developer that gets to a certain level knows that you develop a skill in parsing these. That like, oh, when you have the nested templates, you go all the way down and that's where the error will be and you just skip all that type information. And it's a painful experience
Starting point is 00:35:21 and I have to become an expert in stuff that I don't wanna be an expert in. It's just like experience and I have to become an expert in stuff that I don't want to be an expert in it's just like I want to write code and I want to ship software or like, you know ship libraries and So much of the stuff like I think I mentioned in a different conversation that like the cognitive Load that I have to carry in my head as a C++ dev and so when a tool comes along that like it's you know, like Moses parting the Red Sea and being like you don't need to do any of that stuff. You can just like, you can all of that boilerplate
Starting point is 00:35:49 and not just like documentation, unit tests, you know, and they say don't get it to write your unit tests, but like, do you know what, like writing a Google test, a G test or like a doc test, like there's so much scaffolding and then you can just go and like modify the test for the test case that you want. You want to definitely verify.
Starting point is 00:36:07 You're not just vibe coding your unit test, but do I really want to, oh, what's the macro for the top level thing, and there's the sub case thing, and if you're switching between unit test frameworks, they've all got different macros for stuff. Oh, that's right. You're not allowed to use a lambda inside a macro because it doesn't work that way, so you have to use function objects if you're doing... Anyway, it's just like the AI knows all of that, and you can hold your hand. use a lambda inside a macro because it doesn't work that way. So you have to use function objects if you're doing, anyway, it's just like the AI knows all of that and you just,
Starting point is 00:36:27 you can hold your hand. It's a very lovely experience. And then every once in a while, you do end up having to smash it over the head with a bat, but like that's no more painful than what I was doing before. And so I wonder if people that are coming from like Ruby or Rust that like have much better tool chains and it's less painful,
Starting point is 00:36:44 are they less willing to tolerate when the models don't do well? That was kind of a thought I had in my head. I think that that may be true. I think that that may be true, but then on the other hand, I think that for people writing lower level systems code, large language models tend to do a much poorer job. There's this community called GPU Mode, which is like a GPU kernel hacking community.
Starting point is 00:37:09 And one of the projects they've been working on recently is developing a better training set and model for writing CUDA kernels. And the reason for this is, and this was like as of like three months ago, these evaluations, but they did evaluations of like how did the state of the art models do it writing CUDA kernels. And I don't remember the exact numbers, but I think it was something like for like 70% of the test cases, the state of the art models could not even produce code that would compile. And of the 30% that did compile, only 5% would run.
Starting point is 00:37:52 And of the 5% that would run, only 1% would run fast. So that's a pretty bad failure rate. Only 1% of the tests, the state of the art models could only get the correct code for 1% of the tests. And by correct code, I mean, the reason you're using the GPU is because you want it to go fast. So if the code's slow, you don't care. And so one thing that they started doing
Starting point is 00:38:21 is they run a coding contest on their Discord, where they have coding challenges to write an optimal kernel for something. And the coding challenges are all just an exercise to collect data, to create a curated data set of what good kernels look like, so that they can train or do fine tuning on a model to teach it how to write good kernels. And what I'm getting at here is that if you're writing some lower level systems code, if you're writing like concurrent or parallel code or performance sensitive code, which
Starting point is 00:38:52 is often what you're doing if you're in C++, the models are not necessarily as good at that as they are as like, oh, I want a website for doing x, y, or z. And I think you actually have to do a little bit more hand holding to get them to write correct concurrent code or correct performance-oriented code. So it's interesting because it may mean that, well, yeah, the large-scale fundamentals mean that a lot of the painful parts of doing the C++,
Starting point is 00:39:23 like error messages, et cetera, can be automated away. It may require a little bit more human intervention and somebody who's more senior and who understands what the correct code would look like to be able to get them to generate good results. Yeah, that's fair enough. I definitely have had that experience where when I'm coding CUDA C++ libraries, you really need to... The directions it wants to take you without guidance are typically not great. And even when you spell out what you like, a lot of times they don't do it.
Starting point is 00:39:57 I mean, I just gave that presentation the other day with Thrust where I was like, all right, there might be a single algorithm with the first word and second word of this algorithm name. And it still did not reach for that algorithm. It was like, okay, I'll call these two separate things. So a lot of times, yeah, you need the expertise to know what you want. But if you have that expertise, I still find it more pleasant. And I wonder like if it's a lot of people, they they love the craft of writing code. And so they just have no interest in a tool that is gonna do the writing for them. And I used to think that like, I loved writing code. I know I like it, but like, I don't know,
Starting point is 00:40:38 these tools to me, I feel like maybe that's the thing is I do love and that's maybe it explains why all of my YouTube videos are on these simple little leak code problems where they give you the function definition and the return type and like all You have to do is implement the algorithm. I love doing that stuff. You know what cuz there's no cruft There's no ceremony. You're just it's purely writing the algorithm and I think that's the thing is I love That kind of coding so much of programming and software engineering is not that. It's scaffolding, it's ceremony, it's dealing with errors, it's documentation,
Starting point is 00:41:13 it's unit test framework like setup. It's becoming a second expert in another language called CMake in order to get your build system to work. And I remember before I had my white belt or whatever the second level belt in CMake was. It was so painful trying to get a project to build and to pull in something. Now it's not that bad because I got my blue belt or whatever belt. And also CMake has gotten a lot better in terms of the utilities they provide you. They've got whatever the CPM fetch thing and you just pointed it to GitHub. It's a lot better than it used to be. But just like, I guess my thing here is that like the part that I love about coding, I still get to do. And I can just ignore the other
Starting point is 00:41:54 90%. I don't know. I'm just, I'm trying to figure out why people are so averse to this, what I think is like the best thing to happen in our careers. Well, and I mean, one, like to that point, one, you know, on this question of is is AI going to eat software jobs? It's not like we as an industry, you know, it's not like like software famously, you know, doesn't like ship on time. It's not like we as an industry have a great track record of delivering bug-free code on the deadline that we set out to achieve.
Starting point is 00:42:41 Software projects tend to be over budget, late, and have a lot fewer features than we intended. So we could have a 10x increase in software engineering efficiency. If we had a 10x increase in software engineering efficiency, I don't know that it would necessarily mean that we would have fewer software engineers. It might just like or even fewer software projects. It might just mean that
Starting point is 00:43:10 like, you know, we can deliver higher quality stuff. Yeah, that's fair enough. I mean, what is the when you when you're saying that it came to mind the what's his name? Dunkey videos or whatever is a YouTube comedian that kind of makes funny videos about Video games and he had one on GTA 6 which I think that's the I don't play GTA, but whatever it's one of the rock star games It's very hyped and it's it's been delayed like years in the day That was like the the the punchline of the video is like that the trailer came out and then he was like Whoa, look at look at the graphics. Well, look at the graphics on these graphics and it's just a very beautiful trailer Look at the graphics on these graphics and it's just a very beautiful trailer look at the graphics on this graphics and then he's like unfortunately we'll never be able to play it because it's like delayed until 2026 or whatever yeah anyway so yeah
Starting point is 00:43:57 it's yeah famously we our industry is not great for shipping things on time I mean like where is a Apple's intelligence like they still don't really have AI? Like what's going on? They're working on it. Well and it's it's I mean I think we had to keep in mind that the the models and the technologies that we have today are still pretty rudimentary and limited. I often get frustrated when dealing with Claude in cursor or any model that I'm using in cursor because it's like I'm working with a junior engineer but a junior engineer who never learns. Cursor rules, cursor rules, buddy.
Starting point is 00:44:45 Yeah. But I think that I think that things like cursor roles and all the other equivalents of cursor roles that other frameworks have are a crutch for for like a bigger problem. And the bigger problem is that we need to have AI systems that have a first class, long term memory that things where when I'm working with something, it learns over time. And where it doesn't decohere over time. Because the problem is, you can't have a single session with a model that just
Starting point is 00:45:29 goes on forever because eventually it gets too large and eventually there's too much context or stuff falls out of the context window and things start to decoherent. And OpenAI has some memory system. But as far as I can tell, it's basically where it's just like a tool where when you're in chatting, if you have memory turned on with OpenAI and you're chatting with one of their models
Starting point is 00:45:54 and you tell the model something like, oh, I, you know, we like restaurants that don't have food that's deep fried or too salty. Like, you know, it has a tool where it'll be like store a fact that the user has this preference, or store a fact that the user's name is this, etc. And then it's got a database that has all those facts and it can go and pull those out. And maybe that's what the optimal memory looks like, and maybe that just needs to be matured a little bit. But I suspect that probably the next wave of big innovations that we'll see in models will be some big revolution in memory and in long-term memory.
Starting point is 00:46:38 And that'll give us agents that have greater persistence and that learn from us over time and that sort of grow with us. Yeah, I mean, there's definitely like you learn the patterns of like you said, the first couple of days you were using it and then you kind of got you figured out it's like if you don't use cursor rules, you've probably found out that if you have a repo that has a building.md You shouldn't just ask it to build you should to be safe say build using at building.md And then it never fails with when you tell it where but if you don't tell it Depending on how it's feeling that day, you know, you might have 75% of the time
Starting point is 00:47:20 It does the right thing but usually what it ends up doing when it doesn't do it is it does something that doesn't work and then it greps the repo. Then it finds a building.md and then it goes. But it takes an extra 30 seconds for it to do the grepping and depending on what model you're using, it could be faster or slower. But you learn the golden path of one, just add a cursor rule.
Starting point is 00:47:38 Or if you're not doing that, you just always point it at the building file. And it's true, why does a model after the one time in one session, does that not just like, what's the word, propagate to like every other session that when the first thing it should do behind the scenes when you ask it to build, if you're in a repository, is to go and check if there's a building.md
Starting point is 00:48:00 because obviously like there are the instructions there. I'm sure you could write a cursor rule telling the model to write cursor rules. You could write a cursor rule that says, when I tell you something that's a best practice that you should do in the future, write a cursor rule for that thing. I'm sure you could do something like that.
Starting point is 00:48:22 Or just have some feature built in that whenever you get interrupted by the user Because I imagine like I don't have it set up right now I still am typing all my commands But like I'm pretty sure I'm gonna set up so I can just start talking to this thing and like it's speech to text The other thing I was thinking is like so often my workflow now is because I'm using cloud for Sonic Which it usually takes you know 30 seconds to a couple minutes to maybe even a few minutes, depending on the task and how complex it is.
Starting point is 00:48:48 It takes some time. And so a lot of the times I'm like juggling, I'm juggling multiple things. And I tend not to do to like AI coding sessions. I'll be working on, you know, documentation. Yeah, no, no. Because the thing is, is like, I don't you're definitely not doing it in the same repo, or at least I'm not going to. So it's like.
Starting point is 00:49:06 What you do is you create multiple repos. You create multiple checks out of the repos. There's a couple different techniques for doing this. Yeah, I don't like. I like my workflow. Maybe we should talk about that at some point, what the workflow. But currently, mine is I've got my single repository.
Starting point is 00:49:20 I'm using Cursor, Cloud Force on it. I have the agentic window and I basically, I have everything that auto turns on but Nvidia doesn't let us do the total Yodel-O-Mo, but it'll let you go and search on the web and stuff. And so, you know, I've got it, I've got multiple monitors, you know, so I've got it on my main monitor. I ask it, give it quite a bit of detail.
Starting point is 00:49:41 It sends off to work and then I'm gonna go start like, you know, working on some other task or task, putting them on some stack of things that I wanna do. But at times, you are waiting, it's like you would hope that it was a bit quicker so that I don't need, sometimes it's like five minutes and I'm like, I don't need five minutes.
Starting point is 00:49:59 I'm kinda going back and forth. What I'm thinking is, I spend a lot of time running, listening to podcasts. I would love to just like ambiently, and maybe this is like an unhealthy thing, but like I enjoy doing this stuff so much. I would like to ambiently be like just always talking and like working.
Starting point is 00:50:15 So like if I'm going on an hour run or a two hour run, that's like two hours of lost productivity with GLaD. Where like, if it's, and if I tell it, like take a little bit more time, cause I don't want to every 30 seconds seconds it be interrupting my podcast and my run. But if I say, you know, check back in with me every five minutes, I would be so happy to like not even stop what I'm doing. You give me a pair of like the futuristic meta bands or whatever they're calling meta lenses. Like they're not here yet, but give me a pair of like, you know, Oakley sunglasses where I need a little bit of visual.
Starting point is 00:50:45 I can't just do it based off of whatever, but it gives me like, you know, in my little futuristic holographic thing, I stop running because, you know, I don't want to hit a bike or whatever. But you can imagine like walking around even while you're you're traveling, you're touring and just constantly making checking in and being like, OK, yeah, that's a great direction or no, the next step is to do this. And like how much time does it take to like vocalize and speak that? It's like 10 seconds, 20 seconds. And I just feel like that would be such a cool thing to do is like, you know,
Starting point is 00:51:13 you're, you, you could then technically work and like just go for a walk. Maybe you're not even exercising on a daily basis, but now these tools like make it's like, whoa, I can still continue to absolutely take a meeting or go. It's like we already have a ton of flexibility because we both work remotely. But like I just I think it's going to completely change, you know, the way that we work and I don't know. It's yeah, I agree. And you know, for me, that's one of the really appealing things about it because I I've had all these ergonomic issues for years. I can't spend eight or 10 hours a day at the keyboard.
Starting point is 00:51:49 I can maybe spend three or four hours a day at the keyboard. Sometimes not even that. I've had issues for years. I figured eventually technology would solve this problem, but I thought it was going to be the brain chip. Fun fact, one of my childhood friends was like employee number 10 at Neuralink. So I probably had an in.
Starting point is 00:52:11 I probably could have got him to get one of the chips installed into my brain. But I figured that was going to be the thing, was that I was going to get the brain chip and then be able to interact directly with the computer. But I think it seems like the technological solution is going to be AI. It's going to be that eventually I just, maybe I'm at a keyboard,
Starting point is 00:52:34 but I'm not typing everything out. Maybe I'm typing out one prompter every five minutes or so and letting it work. Or I've been thinking about doing the speech to text thing so that I can go like completely keyboard free. I do I do find that like with the current like cursor UI like I do you know when I'm going and reviewing a diff or something or like I still have to you know go and like touch things or like you know selecting context is a little bit of a hassle so
Starting point is 00:53:05 but I'm sure that there's ways like I'm sure that you could go a lot more compute like keyboard and mouse free if you wanted to which is like really exciting prospect to me so yeah I've been thinking about trying out the speech text although it may drive Ramona crazy to cheer me yelling at the the robot all day. Yeah, I mean, well, that's part of the reason I don't do it right now. But, you know, ideally, at some point we can have a bigger place and separate separate rooms, ideally on separate floors, because she also she works remotely. A lot of the times she has like virtual clinics and whatnot.
Starting point is 00:53:41 But yeah, I mean, when I hear you say that, I worry because Shima gives me a hard time for being what's called a transhumanist, where, you know, we're there's a percentage of the tech bros that are super happy about the future of, you know, integrating and augmenting our intelligence with like hardware. And I posted a tweet a week ago that I asked a chat GPT to summarize like the the timeline of AI like levels because I finished watching a movie called Mountainhead which is a kind of dystopic look at billionaires and how they're destroying the world with tech and anyways it was like we're currently at a and I artificial narrow
Starting point is 00:54:17 intelligence next is a GI which is general after that is SI super intelligence and then the two after that are transhuman Which is like integrating with hardware and then posthuman which was like uploading your cloud to the brain Or uploading your brain to the cloud and anyways when you're talking about like getting a chip in your head So you can talk to the robots. I'm there like being like, oh man, that's just sounds so awesome Like I don't want to be I don't want to be patient zero but I and I know most of our audience is software developers, but I do wonder, like, is the, because I, there was a New York Times daily episode where they were commenting on some stuff.
Starting point is 00:54:52 And they were, their view with people like us is that we are broken morally, and that we're, we don't care about society. And it's like, I definitely care about society, but I also just like, why should we don't care about society. And it's like I definitely care about society. But I also just like why should we put a little box around us of like what we are or like should or should not, you know, like like do I don't know. Maybe I'm not articulating that correctly, but it's just like
Starting point is 00:55:17 it's not that I don't care about society. It's just like we live in a world that's defined by like our history and it's just hard for people to fathom what the future could look like. Yeah. Yeah, I mean, I think generally when we've had great scientific advancements or great advancements in productivity, yeah, like the industrial revolution, there were lots of downsides,
Starting point is 00:55:37 like children losing arms in factories, like stuff like that. It's not all good, but I think on the whole, great advances in technology have improved quality of life for humanity. I mean, I think you look at quality of life today versus 200 years ago, I think it's hard to argue that things aren't better now. You know, that doesn't mean that all the technological advancements have only been used for good or have only had good outcomes. And I don't mean just to suggest that the positive outcomes justify bad stuff that's
Starting point is 00:56:17 happened along the way. And certainly there's lots of ways in which AI is being used or will be used that may harm people. Things like surveillance, it certainly could be something that could become a tool of nefarious actors in various countries that do not enjoy the rights that our two countries have. But I mean, I think on the whole, it's hard to imagine that this won't be good for society unless you're of the perspective
Starting point is 00:56:57 that you believe that P-Doom is likely. And that, you know, it's just hard for me to imagine how that scenario plays out because technology is, at the end of the day, very, very brittle. And it would have to be not so brittle to really be a threat to us. But I also feel like it's almost impossible to control the development of AI. If we as a society decided that we weren't going to build AI because it was too dangerous, I don't know how we could possibly control that. Because it's not like, look at nuclear weapons. Nuclear weapons are something where- I was literally just thinking the same thing in my head.
Starting point is 00:57:45 You like for nuclear weapons, there's a specific set of physical materials that you have to control access to, um, uh, to create them from being created. And like, that's something that like, you can, you can monitor and you can observe, um, uh, and you know, detect through, through satellites to some degree. and detect through satellites to some degree. And even with all that, we've only had like an okay track record of, you know, of keeping new countries from developing nuclear weapons. I mean, look at current events, and look at events from, you know, past 10 to 20 years with North Korea.
Starting point is 00:58:34 AI is not like that. To develop AI, you just need like some dude in their basement who has computers and, okay, maybe you need access to large quantities of compute to really do it. And so maybe there's an argument that that there's a similar like material constraint, but I don't think we're going to, like, you know, burn all the data centers down entirely, because we need them for other things. And if you don't get rid of all the data centers, or control tightly control all the data centers, I don't see how you can, how you can keep people from developing AI. And so it seems like it's a futile effort to try to put regulatory guardrails on it.
Starting point is 00:59:14 And in terms of like technical guardrails of things like aligning models and stuff like that, it seems like we get really mixed results right now. So part of my perspective is like, if the AI is going to kill us, then it's inevitable that we'll develop the AI that will kill us. There's not really much that we as society can do to stop that. So if it's going to turn us all into paper clips, then we're sort of doomed anyways. And if it's not going to, then that's great. And so there's not much for us to do because it's either inevitable doom or we're going to be fine. I mean, I would never word it that way because I'm sure there's some listeners, but I do completely agree. I've always said that we were born arguably at the perfect time because most AI folks,
Starting point is 01:00:08 they say that if the singularity is going to happen in the paperclip, you know, some version of that scenario happens, it's going to be roughly 2050. Some people are saying sooner now because they didn't realize things would accelerate this quickly. But the point being is that like, you know, we're in our 30s, 2050, another 20 years, hit our 50s, you know, we were alive, we got to see it all, and then it could end, but it's like, we are not going to stop that. Like, whether, you know, it's like there's all these different agreements globally, there's always one country that's like, no, no, no, you guys do that, but I'm going to go ahead
Starting point is 01:00:40 and like, you know, make money and we're going to and so that like there's no there's no stopping this and yeah so I kind of agree like if it's gonna happen it's gonna happen but I don't think it's gonna happen and I just think it's gonna like I used to say when I was a bit younger that I actually think I thought it was suboptimal that we the time we got born like imagine if you knew that planes were like like traveling by plane was a thing and you were born like in the 1860s and then like you only at the tail end did you start to see it and then like it became like you weren't able to travel unless if you hopped on a boat that's a whole thing it's like you don't get to go and so now the planes I
Starting point is 01:01:20 have the same view of like technologies that we don't exist at some point you're going to have like shoes that give you the ability to fly. And it's like, I can't believe I live in the time where I don't get to fly with my shoes, you know? It's like, there's some equivalence of that, that is it shoes that make you fly? Is it your ability to teleport from one place to another? It's some crazy thing that we can't fathom
Starting point is 01:01:41 that at the tail end of my life, I'm gonna start to see the beginnings of beginnings of and I'm gonna be like oh damn like it would have been good if I had been born like 50 years later so that I could do that and I feel that like AI potentially is that thing that's gonna unlock all this stuff and you know potentially people are gonna be able to live like way longer and anyway I don't know I I guess part of my thinking here is that I think that, I do think that, like almost any technology, AI and mistakes with AI will likely cause damage
Starting point is 01:02:16 to society and communities. I think it's more likely that this happens not through ill intent, but just through bugs and accidents. In the same way that there have been various disruptions of the internet and of the stock market by bugs in tech. There have been things like the flash crashes from like algorithmic trading companies that
Starting point is 01:02:44 had a bug in their algorithm, trading companies that you know had a bug in their algorithm something like that. I think that for AI to present a real existential threat to us you need a few things. One you would need an actual artificial general intelligence which we're still pretty far from I think. I still think that that's something that we won't see for five to 10 years. You'd need that. And it would need to be a really a persistent thing that is like an ongoing and long running process, right?
Starting point is 01:03:21 And it would need to be incredibly resilient. If it's a thing that lives solely on the internet, I don't know that that is sufficient to present a threat to us because I, I actually might disagree with that. You know, one time in Toronto, payment systems went offline for like four hours and it caused like pandemonium. Like, you know, you couldn't go to a bank to withdraw money. Like you couldn't. Everyone was freaking out. And it was it was like four hours of not having credit card things working.
Starting point is 01:03:59 And like the payment systems were just all down. It was nuts. But there's a big difference between pandemonium and the robot turning this on into paper clips, right? Yes, but you could imagine a world where with deep fake and misinformation and shutting down the internet, it would just lead to absolute anarchy. And it wouldn't be the end of the world, but the amount of damage that it could cause would be like
Starting point is 01:04:31 yeah, a word bigger than massive. I don't need like, you know, completely like I can imagine. I can imagine substantial damage, but not like existential end of human life damage. Oh, yeah. Yeah, not existential. But like, yeah, there'd be millions and millions of people that died. For existential damage, I think you would need to have a persistent artificial general intelligence that has a physical presence, which means that it would need to have robots
Starting point is 01:05:02 or drones or control of physical resources. And it would probably have to have control of those resources in multiple resilient communication mechanisms that can't be disrupted by the various militaries and governments of the world. And that actually seems like that's a pretty high bar to clear to me at least. I just realized like this conversation started off like an hour ago being like, yeah, like, I don't get why people aren't excited about clock four.
Starting point is 01:05:36 And like, you know, I think it's really changing the way the workflow is. Is it because they're worried about jobs? And now we're talking about like, what are the actual parameters that would need to exist in order for the AI, the artificial general or super intelligence to wipe us out? Yeah. I mean, like the...
Starting point is 01:05:56 I do think that like, if you have like a scary like rogue model, some sort of scary persistent rogue model that escaped containment and was like out there on the internet, I, my guess would be that world governments have a greater ability to turn off the pipes of the internet and control things than we might imagine. Yeah, yeah. I mean, that's the case. A lot of countries, they can just turn on and off the internet whenever they want, right? Yeah. I mean, like, you know, it would be pandemonium, but, you know, and then... Yeah, yeah. If you were talking about existential, like, end of life, like humans don't exist,
Starting point is 01:06:42 yeah, that's definitely not what would happen. There would be a bunch of folks that survived. It wouldn't be a good situation, but like, yeah, we would be able to rebuild for some definition of the word rebuild. Be sure to check these show notes, either in your podcast app or at adspthepodcast.com for links to anything we mentioned in today's episode, as well as a link to a GitHub discussion
Starting point is 01:07:00 where you can leave thoughts, comments, and questions. Thanks for listening. We hope you enjoyed and have a great day.

There aren't comments yet for this episode. Click on any sentence in the transcript to leave a comment.