Latest / Coffee with Developers / WeAreDevelopers LIVE - Your Brain Is Lying to You: Cognitive Biases in Development - Kate Astrid

Transcript

WeAreDevelopers LIVE - Your Brain Is Lying to You: Cognitive Biases in Development - Kate Astrid · Coffee with Developers

• 1:28
  SPEAKER_02: Hello and welcome.
• 1:30
  SPEAKER_02: And why is Microsoft shared already?
• 1:32
  SPEAKER_02: Fair enough.
• 1:32
  SPEAKER_02: Cool.
• 1:33
  SPEAKER_02: It's never gonna happen that it's that nothing happens wrongly, right?
• 1:37
  SPEAKER_03: It's just good fun.
• 1:38
  SPEAKER_03: So that's live for you.
• 1:40
  SPEAKER_03: Isn't it?
• 1:40
  SPEAKER_03: That's the way it goes.
• 1:41
  SPEAKER_03: I feel like it's kind of part of it.
• 1:42
  SPEAKER_03: You know, it should be built into the name.
• 1:44
  SPEAKER_03: If it's live, things will go awry.
• 1:46
  SPEAKER_03: But there you go.
• 1:47
  SPEAKER_02: Well, it's the same like live coding and like or or like uh deploying to live rather than like deploying to tests and these kind of things.
• 1:54
  SPEAKER_02: So it never should do that.
• 1:56
  SPEAKER_02: I'm in the middle of actually re-homing my blog, and I just wrote about it, and it's just pretty hilarious to see how much I accustom uh I accumulated over the last years.
• 2:05
  SPEAKER_02: I mean, 21 years of blogging, and it's like a 2.7 gig file of like static pages now, and now I I deployed it to GitHub pages and it's like an 840-megabyte GitHub file.
• 2:17
  SPEAKER_02: So um, yeah, it's gonna be interesting to see not to make all these things work.
• 2:22
  SPEAKER_02: I realize I got like internally 8,300 links, and so far I found 500 broken ones, but that's basically because most of the services that I point to don't exist any longer, which is kind of uh disappointing to see how much that the web is actually dying and uh yeah, and nobody cares about that.
• 2:41
  SPEAKER_03: Well, you know, I think it'd also be interesting.
• 2:43
  SPEAKER_03: I'd like to see some of the um kind of top like headline numbers from that, because I think you've written, you know, you know so many languages, you've worked in so many different kind of environments that it'd be interesting to see if like the trends happened, you know, in terms of when did when when frameworks came in, did the kind of topic uh the topics that you were talking about change in that maybe there was less kind of platform HTML stuff if if it was kind of all being kind of wrapped into React hooks and things like this, you know.
• 3:05
  SPEAKER_03: Um just being interesting to see what the kind of trends were over the years.
• 3:08
  SPEAKER_03: 21 years, fair play, Chris.
• 3:10
  SPEAKER_03: That's impressive.
• 3:11
  SPEAKER_02: Uh I've yet to use React though.
• 3:14
  SPEAKER_02: I think that's the but that that's the main thing.
• 3:16
  SPEAKER_02: So that basically that will not happen that I've written about it, except for probably complain about it.
• 3:21
  SPEAKER_02: But yeah, what can you do?
• 3:22
  SPEAKER_02: I mean, it's my blog and I I I um it's my party and I cry if I want to, so that's not really the issue.
• 3:27
  SPEAKER_02: You keep blocking out right now again as well.
• 3:30
  SPEAKER_02: Do you have some some uh drive thing happening or something?
• 3:33
  SPEAKER_03: Or I can shut absolutely everything, which is always a good uh good good idea, isn't it, anyway?
• 3:40
  SPEAKER_03: But yeah, we never know.
• 3:40
  SPEAKER_03: It's always something kind of in the background, isn't it?
• 3:42
  SPEAKER_03: That we found was it clawed or something for you once was kind of causing your screen to stop sharing?
• 3:47
  SPEAKER_03: And so yeah, it's all these kind of background tasks, but fingers crossed, most mostly the ones that could kind of mess with me have gone.
• 3:53
  SPEAKER_03: We'll see.
• 3:54
  SPEAKER_02: Bloody, bloody clawed, always the same issue, isn't it?
• 3:57
  SPEAKER_02: Yeah, what can you do?
• 3:58
  SPEAKER_02: Cool.
• 3:59
  SPEAKER_02: Uh so yeah, welcome to We Are Developers Live.
• 4:01
  SPEAKER_02: We haven't even said that one yet.
• 4:03
  SPEAKER_02: Uh, it's gonna be uh 28th of October edition, and um yeah, tomorrow I have a coffee with developers with Werner Fogels from the CTO from Amazon.
• 4:12
  SPEAKER_02: So I'm preparing a few interesting questions for that one.
• 4:15
  SPEAKER_02: So it's gonna be interesting to see what else we have and what else is coming down the pipeline.
• 4:20
  SPEAKER_02: And then, of course, we're getting ready for India.
• 4:23
  SPEAKER_02: And uh yeah, well, 21st we're supposed to be flying, and so that's not that's four weeks off, so not even less than four weeks off.
• 4:31
  SPEAKER_02: So let's see what we can do.
• 4:32
  SPEAKER_02: Uh we got the visas, we got everything sorted, so hopefully it's gonna work out.
• 4:36
  SPEAKER_02: And we have a guest today, of course, as well.
• 4:38
  SPEAKER_02: And the guest today is Kate Astrid, uh product engineer at Cogram.
• 4:42
  SPEAKER_02: What does Cogram do?
• 4:45
  SPEAKER_01: Hello, hello guys, it's so nice to be here.
• 4:47
  SPEAKER_01: Thank you so much for having me.
• 4:48
  SPEAKER_01: So, Kogram is actually working on a platform for architectures, engineers, and construction builders.
• 4:54
  SPEAKER_01: And we create the platform that helps them to work with meetings, projects, and basically helping them unify all of the work in one place.
• 5:06
  SPEAKER_02: So you work with proper architects, not software architects.
• 5:09
  SPEAKER_01: Yes, yes, with real people that build real stuff.
• 5:13
  SPEAKER_01: Most of our customers are actually in the States, and I was in the States a month ago and I saw some buildings that they made, and it was incredibly fascinating to see them in person, you know.
• 5:24
  SPEAKER_02: And engineers, like proper structural engineers as well.
• 5:28
  SPEAKER_02: So, yeah.
• 5:28
  SPEAKER_05: Yes.
• 5:29
  SPEAKER_02: That's interesting.
• 5:30
  SPEAKER_02: I love that we occupied all these different job titles from other markets in in IT, and nobody ever questioned that if that's okay or not.
• 5:38
  SPEAKER_02: Like, I I think I mean you cannot call yourself a doctor, but you can said uh help call yourself a healing practitioner, for example.
• 5:46
  SPEAKER_02: You don't need an education for that.
• 5:48
  SPEAKER_02: It's not it's not a safe uh it's not a copyrighted term or like a term where you can pretend.
• 5:53
  SPEAKER_02: And I remember uh I always I always had arguments with my wife about this because I used to be a journalist and I went to journalism school.
• 6:00
  SPEAKER_02: I didn't study journalism, I only worked on the job in the radio and had like a three-month course to basically become a radio journalist.
• 6:08
  SPEAKER_02: And she always like, whenever I introduce myself to people, and they're like, Don't say you're a journalist because you haven't studied it, like it's basically showing off.
• 6:15
  SPEAKER_02: And I'm like, But but I did that course and I basically got certificate and everything.
• 6:19
  SPEAKER_02: But it seems like uh tricky that some jobs outside of our realm uh seem to be a thing that you need to actually uh uh apply for and work hard for.
• 6:30
  SPEAKER_02: And and uh with IT, I always like, what do you want to call yourself in your job title?
• 6:36
  SPEAKER_02: And we said that before, it's dangerous to actually call yourself the coolest ninja, uh whatever cool guy uh uh in your job title because it will it's not comparable.
• 6:47
  SPEAKER_02: And when people look for somebody to hire, they're not looking for like the the quirky name that your startup gave you this month.
• 6:55
  SPEAKER_00: Yeah, that's true.
• 6:56
  SPEAKER_00: That's true.
• 6:58
  SPEAKER_03: So it doesn't always tell people what it is.
• 7:01
  SPEAKER_03: It doesn't always tell people, it doesn't always tell people really what it is as well, you know.
• 7:04
  SPEAKER_03: Like I think some of these ones that are the kind of really cool kind of hip names that you're kind of hearing about that are just um they're they're they're almost like social media names, you know.
• 7:12
  SPEAKER_03: They're they're kind of things that you then uh kind of know the person as, you know, rather than sort of actually what the job is.
• 7:19
  SPEAKER_03: So it ends up becoming and I wonder if maybe this is actually kind of uh do we think perhaps we're gonna see more of this?
• 7:26
  SPEAKER_03: As as we're kind of seeing, you know, the front end becomes full stack, and then now we've got sort of everyone expected to do everything, will we see kind of these job titles that uh are less uh specific becoming more common?
• 7:38
  SPEAKER_01: Well the new thing is forward.
• 7:40
  SPEAKER_01: Yeah, forward deploy engineer.
• 7:42
  SPEAKER_01: I think we wanted to stay the same, right?
• 7:44
  SPEAKER_01: Yeah, it's something completely new, and I was like, who are these people?
• 7:48
  SPEAKER_01: And then I basically find out that this is kind of the same what I'm already doing, you know, but just with different titles.
• 7:54
  SPEAKER_00: Okay, guys.
• 7:56
  SPEAKER_02: So yeah, it's basically a consultant being on site working with the client that uh and that used to be a job that we had in the 80s and 70s, and so that basically I was not being a consultant because I went to companies and they already knew what was broken, they just basically had their engine their my my managers not uh listen to their engineers, but they had to get somebody outside in who's expensive.
• 8:19
  SPEAKER_02: And if that person says that, then basically company would do something about it.
• 8:23
  SPEAKER_02: So uh it was it was kind of a depressing job, though, because you always went there to to catch uh uh to to douse fires and basically sort out things that are broken.
• 8:33
  SPEAKER_02: Nobody liked you coming in because they basically said, like, okay, they they hired this guy to take my job away.
• 8:39
  SPEAKER_02: Now we have that with Asians, so that's really not that much of an issue.
• 8:42
  SPEAKER_02: But yeah, forward deployed engineer always sounded like super complex.
• 8:46
  SPEAKER_02: I have the problem right now when I fill in forms, like I we just filled it out easier for India, and I'm now vice president developer relations, and that's first of all too long for most of the forms.
• 8:57
  SPEAKER_02: And if you just tell people I'm a vice president, what the hell does that mean?
• 9:00
  SPEAKER_02: Like, do you run a country?
• 9:02
  SPEAKER_02: Do you like you know I'm a vice president?
• 9:05
  SPEAKER_02: Okay, fair enough.
• 9:07
  SPEAKER_03: Like, you know, yeah.
• 9:08
  SPEAKER_03: I feel like if you're choosing though between VP and DevRel, I would probably include the VP bit because that sounds like you're gonna get some special treatment.
• 9:15
  SPEAKER_03: Whereas DevRel, you're gonna be stood at the border trying to explain to somebody what the developer relations is, and you're gonna be saying, I have enough problems trying to explain it to developers, let alone kind of border patrol, you know.
• 9:25
  SPEAKER_02: I always put software engineer in there because for 20 years I had to explain what a software engineer does, and now people know it.
• 9:32
  SPEAKER_02: So that's basically that was fun when I worked with Tim Berners-Lee, like the inventor of the World Wide Web and the get a person, and his job, his pre his business card basically says web developer.
• 9:44
  SPEAKER_03: And I love that because nicely done, understated, very nice.
• 9:50
  SPEAKER_02: So we're three we're three days away from Halloween, so I thought we make it a bit spooky.
• 9:55
  SPEAKER_02: This is like our spooky edition, and I I'm I you I mean, I have to say, like hard uh like brace yourself because this is gonna get really spooky.
• 10:15
  SPEAKER_02: So I mean, if if ghosts are unreal, try to explain that.
• 10:20
  SPEAKER_02: Like, that's a tricky thing.
• 10:22
  SPEAKER_02: Yeah, I think the bigger problem is that everybody expects like the youth to riot on the weekend now because there's lots of in Germany, especially there, there's lots of people that said, like, we're gonna go on the streets and basically talk about that we get affordable rent, that we get affordable uh um income and these kind of things, and basically everybody looked at what for happened in France, and now everybody's worrying if that's gonna happen on uh on Halloween this year as well, which brings me kind of it feels like the the movie The Crow, where they had like the night before where all this the illegal stuff happened before Christmas, so the hello all Hallows Eve kind of thing.
• 10:55
  SPEAKER_02: Yeah, we'll see how it's happened.
• 10:57
  SPEAKER_02: It's Germany, it's probably people filling out forms to riot or something, or uh or apply for a riot badge, or I don't know.
• 11:04
  SPEAKER_02: We'll see.
• 11:05
  SPEAKER_02: License.
• 11:06
  SPEAKER_02: Yeah.
• 11:07
  SPEAKER_02: Cool.
• 11:08
  SPEAKER_02: So uh let's take a look at some of the bits that I wanted to talk about before we go into the deep the deeper topic that Kate uh has a specialty in, which is cognitive biases, uh, which is something we all have, and uh it's actually fascinating to see how much better you become in teaching, learning, and understanding people when you face your biases as well.
• 11:30
  SPEAKER_02: But let's get to that further.
• 11:32
  SPEAKER_02: Uh in depressing news, which I found really annoying, is that uh Microsoft is shutting down the garage, uh or garage, as the Americans say, uh, which is a program behind its hackathon and makerspaces.
• 11:44
  SPEAKER_02: So we had garages everywhere in the planet, and well, we don't work for Microsoft anymore.
• 11:50
  SPEAKER_02: Uh the Microsoft had garages everywhere on the planet, and they were wonderful places because they were not only for like code hackathons, but you had people that worked with Arduinos that built little robots.
• 12:00
  SPEAKER_02: We had like uh um uh our own basically we grew our own salad in the one in the main office in in Redmond, and uh it was always a great place that you could hire to uh to basically do a meetup and do these kind of things.
• 12:15
  SPEAKER_02: Uh, I'm not sure what the reason was, I guess money.
• 12:19
  SPEAKER_02: Uh, but it's kind of sad that these things go away.
• 12:22
  SPEAKER_02: And I mean, many other companies have done the same thing as well.
• 12:25
  SPEAKER_02: Like, there's a lot of places that did develop community outreach, and uh a lot of this stuff has been underfunded and basically uh turned down, which is kind of a shame.
• 12:35
  SPEAKER_02: I think the uh the generators still exist.
• 12:39
  SPEAKER_02: Microsoft Generator was another brand that they had for like co-working spaces.
• 12:43
  SPEAKER_02: Uh but yeah, the the mix of especially right now with AI uh making all writing all the code, I thought that going into the maker world would be much more a thing that we do right now, that we mix hardware and software much more than we did before.
• 12:58
  SPEAKER_02: But who am I to actually say that?
• 13:01
  SPEAKER_03: What about for you for both of you though?
• 13:03
  SPEAKER_03: As in do you because I I sort of would say that I have had the thought that a few times, you know, when you're saying about things like Arduino, I looked into that years ago, I was very, you know, I was I wasn't even really writing code at all, so I was kind of it was not where I was gonna start, you know.
• 13:16
  SPEAKER_03: But there there is something about when now we've got this kind of idea that we can, you know, at least semi-vibecode things or even kind of do small scale things.
• 13:24
  SPEAKER_03: Uh have you thought about doing that?
• 13:25
  SPEAKER_03: Like making, you know, I've seen people hacking their TVs because it's running on um Android, and people are using their Raspberry Pi to, I don't know, uh pre-program the lights on their Christmas tree and things like this.
• 13:36
  SPEAKER_03: And you're already seeing these kind of ideas that somebody who maybe works on the web wouldn't have gone and done that, uh most likely.
• 13:43
  SPEAKER_03: Uh do either of you kind of have that?
• 13:45
  SPEAKER_03: Do you feel like there's some of your ideas are now accessible in a way that they weren't before?
• 13:50
  SPEAKER_01: Yeah, of course.
• 13:50
  SPEAKER_01: There are lots of things that you can do uh without having any uh specific knowledge for that.
• 13:56
  SPEAKER_01: A friend of mine recently hacked his uh freege, you know, that uh fridges that have a screen and with some figures there and people there like AI stop with you.
• 14:06
  SPEAKER_01: So he hacked that in the friend of mine apartment, and now this fridge is sharing some additional news every day to a friend of mine.
• 14:15
  SPEAKER_01: So I think it's very funny.
• 14:18
  SPEAKER_01: So yeah, you can do lots of different stuff.
• 14:22
  SPEAKER_02: Well, that was especially important as Samsung was actually showing you ads on your own T uh on your own fridge.
• 14:29
  SPEAKER_02: Basically, you bought like a 3,000 euro fridge and then they showed you ads every morning.
• 14:33
  SPEAKER_02: And I'm like, this is just depressing.
• 14:35
  SPEAKER_02: It's a bit like same with the smartphone, uh, smart TVs.
• 14:38
  SPEAKER_02: I mean, I'm I'm super annoyed that my Fire Stick gives me like 90% ads and 10% of the stuff that I paid for.
• 14:44
  SPEAKER_02: And uh I'm I'm not surprised that people start hacking this.
• 14:48
  SPEAKER_02: I've done a lot of Arduino stuff in the past.
• 14:50
  SPEAKER_02: Uh I still have to set up my my uh my Spark GTX.
• 14:55
  SPEAKER_02: Uh and uh I will hack with that one a lot as well.
• 14:58
  SPEAKER_02: But uh it's gonna be it's interesting that this is a different world.
• 15:02
  SPEAKER_02: I think the other bit is that people were worried about bricking their stuff, and now they're not as worried about that anymore because AI can undo the things that they've done before as well.
• 15:13
  SPEAKER_02: But that was always the thing.
• 15:14
  SPEAKER_02: Like, I mean, back in the uh in the early days of of like hardware hacking, I remember when people bought BMWs and actually put Linux on them and then couldn't drive their car anymore and had to spend like another 12,000 euro to get it fixed by the by BMW itself.
• 15:28
  SPEAKER_02: And but yeah, good luck, uh fun thing.
• 15:30
  SPEAKER_02: You had like a command line for like 10 minutes on your on your car.
• 15:33
  SPEAKER_02: Good thing.
• 15:34
  SPEAKER_02: Uh but hardware, uh hardware emulation as well has become a much better with the with AI space.
• 15:43
  SPEAKER_02: I see this in the retrocomputing world that I work in.
• 15:46
  SPEAKER_02: Like, I mean, if I do Commodore 64s and stuff, we had emulators for quite a while, but now it's actually all part of the of the editor of uh there's VS Code plugins that simulate the machines inside VS Code for you, and that has all been written uh by not white-coded, but like generative coded in the uh because it's much easier that way for you.
• 16:05
  SPEAKER_02: That also floods the market to a degree as well.
• 16:08
  SPEAKER_02: That's the other problem.
• 16:09
  SPEAKER_02: There will be a lot of tools that we don't need to have.
• 16:12
  SPEAKER_02: I actually feel it's interesting that uh Meta just released their uh hardware thing that like I don't even know what it is.
• 16:19
  SPEAKER_02: I think it's like a little Tamagotchi kind of thing uh that supposedly you take with you.
• 16:24
  SPEAKER_02: And I love that all of those failed spectacularly in the past.
• 16:28
  SPEAKER_02: Like there were always like these like pendants that would that would read uh listen uh to the audience and then give you AI answers, and nobody gave a crap after two months.
• 16:38
  SPEAKER_02: And I don't know if Meta is going to be the one that actually turns that tide around.
• 16:43
  SPEAKER_02: But when you think about it, the same with like smart glasses, when even Apple failed, when everybody said, like, okay, uh Google Glass, nobody wants to have Microsoft's uh holo lens is far too expensive, and then Apple is cutting into that.
• 16:56
  SPEAKER_02: And I remember at our event, uh the uh then the CTO of GitHub gave out a uh uh uh a what was it called?
• 17:04
  SPEAKER_02: Apple lens, Apple Vision Pro.
• 17:08
  SPEAKER_02: Apple Vision Pro.
• 17:09
  SPEAKER_02: He gave out one of those as a price, and then like three months later, nobody in Europe could actually use them anymore.
• 17:14
  SPEAKER_02: So this is like this is a tricky thing because they get expensive as soon as hardware is involved.
• 17:20
  SPEAKER_03: I don't know if you can fake it as well, you know.
• 17:23
  SPEAKER_03: I feel like you can't force these things on people, like the the the popularity has to be organic and it has to actually be something that's really useful.
• 17:30
  SPEAKER_03: I think there's a reason why the iPhone became what it is, right?
• 17:33
  SPEAKER_03: It wasn't just that it was kind of something new and hot and sold by Apple.
• 17:36
  SPEAKER_03: It was actually giving you features that were gonna change I hate to use that phrase, change the the game.
• 17:41
  SPEAKER_03: But um, you know, I feel like with new technology you can throw interesting things at the wall, but actually widespread adoption is just gonna happen or it's not, and it's I don't think it's something that the uh the company can really force upon you.
• 17:53
  SPEAKER_03: So it'd be interesting to see though if it does uh this Tamagotchi thing becomes uh popular.
• 17:57
  SPEAKER_03: I I don't see it personally, but there you go.
• 18:00
  SPEAKER_01: Yeah, speaking about Apple, I am now I am now very curious about this uh foldable iPhones that uh Apple recently released because before it was only like Samsung idea, right?
• 18:11
  SPEAKER_01: And only some people who I don't know wanted some toy, they would buy it because I cannot find any use cases why Apple would release that.
• 18:20
  SPEAKER_01: Do you guys know about any possible use cases for uh for foldable iPhones?
• 18:26
  SPEAKER_02: Well, I mean, uh Microsoft had the duo before, and that was really good for Teams because you would have the chat on one side and the screen on the other.
• 18:34
  SPEAKER_02: So basically, that was the I like that one, but it had a physical hinge.
• 18:37
  SPEAKER_02: And I'm always worried about screens that fold, and and there's also reports right now that actually the first iPhones are actually being delayed because people are already breaking their their screens uh because of hinges, and you're like, Well, nobody should have could have seen that coming, right?
• 18:52
  SPEAKER_02: Like, you should not fold glass, it really makes no sense.
• 18:56
  SPEAKER_02: But um, I mean, I've seen people use the HTC ones, the little ones, and they were excited about that one.
• 19:02
  SPEAKER_02: It would be fun to have one that folds out into a big screen like eight times, so you can actually wrap it up in a cafe and put it up there.
• 19:09
  SPEAKER_02: But um, yeah, I mean, yeah, I think it's uh as much as we we we keep uh uh uh uh praising jobs too much, uh and John Ive.
• 19:20
  SPEAKER_02: I think after they left, Apple is now scrambling to actually try to actually uh support a market that wants things, and it's a classic case of like Apple was always good about not building features, Apple was good about like finding the real solutions that people need and then release them and not just add more and more features.
• 19:38
  SPEAKER_02: I mean, Samsung phones always had like 50 features that the iPhone didn't have, and hope that by putting more features in, you sell the thing more.
• 19:45
  SPEAKER_02: And that didn't work out because every feature is a decision, and every feature is something that needs to be maintained and understood by user.
• 19:52
  SPEAKER_02: It's a bit like UX uh interface design as well.
• 19:55
  SPEAKER_02: The more the more bits you put in the screen, the more cognitive load the user has to understand those things because we just can't block them out.
• 20:02
  SPEAKER_02: So Apple was really good at just doing one thing well and doing one thing easy, and that's why it was succeeding compared to the other smartphones.
• 20:10
  SPEAKER_02: But now Apple is caving in, like that they're coming up with touch screens now, which they said for years they're not gonna do, except for like the iPad, but there's gonna be MacBooks with touch screens now as well, and there's this foldable thing as well.
• 20:23
  SPEAKER_02: So I don't know.
• 20:24
  SPEAKER_02: I mean, they have more money than God.
• 20:26
  SPEAKER_02: I I I I I guess it's the the classic case of like you have to keep growing or keep growing, or you're seen as a failure.
• 20:35
  SPEAKER_02: Talking of talking about failure.
• 20:37
  SPEAKER_02: Uh, I mean, the last few weeks the the internet was down, and it was basically because of PHP.
• 20:43
  SPEAKER_02: And it's so funny as a as a very, very dedicated PHP developer, how many people started talking about it?
• 20:48
  SPEAKER_02: And I saw it in the news on television, like people talked about PHP out of a sudden.
• 20:53
  SPEAKER_02: I haven't uh uh I should email Rasmus because I worked with him before Rasmus Leadov, the inventor of it.
• 20:59
  SPEAKER_02: What what I can't even imagine what uh what he had to go through.
• 21:03
  SPEAKER_02: But the uh uh looking into it, it's not surprising that it that it became such a thing because yeah, the PHP Foundation has one full-time staff member, and that's just amazing, you know.
• 21:16
  SPEAKER_02: And uh it's basically uh it's it's uh the other one, uh the Atlantic put it differently.
• 21:21
  SPEAKER_02: It basically said, like, if anything happens to this guy, we're all screwed.
• 21:25
  SPEAKER_02: And uh, I think that's something that we should be considering because there's so many open source tools out there that run the internet, and uh nobody gave them any love or gave them any consideration or gave them any uh any funding in the last few years.
• 21:41
  SPEAKER_02: And yeah, of course, then something like that will happen.
• 21:44
  SPEAKER_02: I'm not surprised.
• 21:45
  SPEAKER_02: I'm actually surprised how fast it was fixed.
• 21:47
  SPEAKER_02: It was really, really good for doing that.
• 21:50
  SPEAKER_02: And I mean, PHP to shell was a big issue, but then there was also uh, I mean, we we we had heart bleed before.
• 21:57
  SPEAKER_02: Uh I mean, it was a double whammy.
• 21:59
  SPEAKER_02: We had the PHP.
• 22:00
  SPEAKER_02: HP problem, and then we had the uh the certificate issue in India, where like the root certificates of the internet basically uh failed because they had a password of one, two, three, four or something like that.
• 22:11
  SPEAKER_02: And uh so yeah, the whole internet kind of went away, and of course, the AI bros are all like, oh so now we gotta get rid of the open internet because it's insecure.
• 22:20
  SPEAKER_02: No, it's just let's let's pump things, that will be good.
• 22:24
  SPEAKER_03: How to take the wrong, uh, the wrong conclusion from that story as well, you know, that we should shut down the open web.
• 22:29
  SPEAKER_03: Like that is totally the wrong conclusion.
• 22:31
  SPEAKER_03: I mean, the thing the thing that when you said Chris, like these things have been um, I can't remember the exact phrase you used, but you said like these haven't been given enough love and attention and funding, you know, for so long.
• 22:40
  SPEAKER_03: The the the individual who's actually doing the maintaining and doing the fix, all of that applies directly to him because when you read the um the article, it says in there that he he does he does this obviously on a volunteering basis and that it's uh in addition to his kind of um full-time job and that he is sustained by kind of uh small donations.
• 22:59
  SPEAKER_03: And I'm just really hoping that like he doesn't have to think about it anymore, you know, that actually that the donations hopefully that people would have kind of recognized his work and said, right, let's let's sort him out.
• 23:09
  SPEAKER_03: And at least, at the bare least, if he wants to do it, you know, give him the financial kind of means to do it full-time, because I think we all now kind of see that we need him.
• 23:18
  SPEAKER_03: Um, and we need other people, we need him to train other people, you know, and to kind of make it so it's not all reliant on him, but that's not the conclusion you should take, like, let's close the open internet person.
• 23:28
  SPEAKER_02: But that's also exactly the wrong answer, as well.
• 23:30
  SPEAKER_02: Like, we don't need him, we need him to have more people to to to fall back on.
• 23:36
  SPEAKER_02: And that's another thing that basically like companies do that as well.
• 23:39
  SPEAKER_02: Like, oh, we got that one full stack developer that can do everything, and you're like, no, let's have four developers to basically do the thing.
• 23:46
  SPEAKER_02: We can afford them.
• 23:47
  SPEAKER_02: Like, if you if you saved so much money by being AI uh native now, then you can afford hiring more people, right?
• 23:56
  SPEAKER_02: Yeah, but you never hear never hear that you never hear that answer, though.
• 24:01
  SPEAKER_02: That's the problem with it.
• 24:03
  SPEAKER_02: Yeah, but I mean uh OpenAI basically said that as well that they're now no longer trained its AI models on the live internet, citing significant risks because of that.
• 24:11
  SPEAKER_02: So they now basically, I mean, it's kind of ridiculous because now they're caching and uh training it on the cached versions, which is something they should have probably done from the very beginning rather than pinging the open internet to death.
• 24:27
  SPEAKER_03: Yeah, it does feel like an afterthought, doesn't it?
• 24:29
  SPEAKER_03: It it does fit and I wonder as well with with this, like how much of it is um, I don't know, tied up with other stuff like intellectual property and kind of the potential, you know, the if you imagine that the um the amount of data that's been you know scraped and indexed, that that there's a lot of risk there, right, for the the the company in doing that, that if if there's some big landmark court case or anything like that that goes um uh one direction against them, you can kind of understand why they might want to mitigate that like going forward, maybe not just constantly kind of scraping everything on the on the live internet.
• 25:00
  SPEAKER_03: Um I think also just to finish up on this, there is something that's really interesting actually in there about how um kind of when they're when they're saying about how the that they're gonna sort of train them in sandboxes, uh pretty much um again, people that know a lot more about this, but it seems to be pretty much consensus that they're just saying you can't do that.
• 25:16
  SPEAKER_03: Uh it won't work.
• 25:17
  SPEAKER_03: Um and so it's just seems a bit of a non-starter that there doesn't seem to be a lot of belief from the kind of development community or the or the tech community that this is actually possible or that actually it would be feasible financially.
• 25:28
  SPEAKER_03: They're saying it would cost them a lot to uh effectively have somebody in charge of keeping it in in the sandbox.
• 25:34
  SPEAKER_03: Um so I don't know, don't know what this really means quite just yet.
• 25:39
  SPEAKER_02: Talk about spooky version of the live show.
• 25:42
  SPEAKER_02: You said one direction.
• 25:43
  SPEAKER_02: What's wrong with you?
• 25:44
  SPEAKER_03: I did, I did, I noticed that as well.
• 25:46
  SPEAKER_03: Yeah, it popped into my head as well.
• 25:48
  SPEAKER_02: Yeah, yeah.
• 25:49
  SPEAKER_03: Moving on, moving on.
• 25:52
  SPEAKER_02: Yeah, another big thing was uh uh uh was popcorn time.
• 25:56
  SPEAKER_02: I don't know if you've used it or you heard of it.
• 25:58
  SPEAKER_02: It was basically an interface on top of streaming uh uh sites, uh, but also illegal streaming sites that allowed you to actually watch movies uh uh from Plex servers and from torrents and all kinds of stuff.
• 26:10
  SPEAKER_02: And that one has been taken off by a huge botnet and uh was a massive security uh scare as well the last few days.
• 26:18
  SPEAKER_02: And basically, there's two million updates remotely disable popcorn time update.
• 26:22
  SPEAKER_02: So um, if you use popcorn time, you're probably your machine is screwed right now.
• 26:27
  SPEAKER_02: And again, I'm always fascinated when people are confused about this, when they're like, Oh, I used something half legal and I wondered why something went wrong with my computer.
• 26:38
  SPEAKER_02: Like yeah.
• 26:41
  SPEAKER_00: Surprise, surprise.
• 26:44
  SPEAKER_00: Yeah, surprise, surprise, yeah.
• 26:46
  SPEAKER_03: I mean, I I do kind of wonder though, like let's be honest, right?
• 26:50
  SPEAKER_03: I mean, I I know that I was kind of when I started kind of getting into everything from building websites to just being chronically online, um, when I was sort of 13, 14.
• 26:57
  SPEAKER_03: Um, the amount of time that I suppose the amount of times I did that where I would have a computer, I would download some kind of.exe file, not not really knowing what it was.
• 27:06
  SPEAKER_03: And then it was a matter of time until I had a problem with my computer, you know, and it and it wasn't always like a direct thing.
• 27:12
  SPEAKER_03: I'm sure it was not me not understanding what those vulnerabilities were, that I'm sure a lot of the time they were these kind of drip feed, kind of long-term ones, you know, but it would I'm sure, I'm sure that some of those torrents or things that I downloaded, especially the uh uh the suite that I won't kind of say the name of the company, but it's lots of graphics and video kind of uh software.
• 27:33
  SPEAKER_03: I mean, as I've said, um I may or may not have in the past, as a student, kind of used those ones that you'd find online, and they were nearly always very dodgy.
• 27:41
  SPEAKER_03: Not just because I don't know for what reason, but that whatever group company, whatever it was that would hack these things, would always have a quite a scary looking kind of logo of like a skull, like a bright green neon skull or something like this, where it made the software look extra dodgy, but at that point I just needed the software, so it kind of shows you, isn't it, like I was willing to take a massive risk from something I didn't trust because I just wanted the free stuff.
• 28:07
  SPEAKER_03: Uh big issue.
• 28:10
  SPEAKER_01: Yeah, but then you have this and then you have this pop-up with like uh yeah, the skull and the text that please send me all your money, otherwise your laptop is completely blocked, and you have lots of viruses that will erase all of your data.
• 28:24
  SPEAKER_01: I had that one.
• 28:26
  SPEAKER_02: Yeah, yeah, yeah.
• 28:27
  SPEAKER_02: We've all been there.
• 28:28
  SPEAKER_02: Yeah, I think my favorite were people that downloaded cracked versions of antivirus software, and you're like, um no.
• 28:38
  SPEAKER_02: That is like the worst idea, but okay, fair enough.
• 28:42
  SPEAKER_02: Yeah.
• 28:42
  SPEAKER_02: Um, the other thing was also interesting to learn was that uh AI models cost a lot more when you don't speak English.
• 28:50
  SPEAKER_02: So basically, I I always heard that Chinese is actually more effective than English, but apparently there's new research now that says it's not.
• 28:58
  SPEAKER_02: I don't even know.
• 28:59
  SPEAKER_02: Uh uh, I mean, German is longer, so I guess that's another issue as well.
• 29:03
  SPEAKER_02: That uh and Dutch is weird, so that might be another issue, but just great.
• 29:12
  SPEAKER_01: But actually, what is interesting, like right now I am in Georgia, and Georgia, the country, has its own alphabet, which looks like this I don't know, like uh branches of uh grapes.
• 29:25
  SPEAKER_01: No, no, it's not Iridi, it's like hieroglyphical somehow, you know.
• 29:29
  SPEAKER_01: You can you can you can Google it, it looks fantastic, very beautiful, and it's unique for this specific country.
• 29:35
  SPEAKER_01: And I read a very interesting research about this.
• 29:38
  SPEAKER_01: So if you do something in Georgian in terms of AI, it cost you like 10 times more in tokens than in English.
• 29:46
  SPEAKER_01: So there was a research about one, I don't remember the sentence per se, but the sentence was like somehow very simple, and it cost uh 15 tokens in English and 156 tokens in Georgian just because of the language, how it's being compiled and all of that.
• 30:02
  SPEAKER_01: So it's crazy.
• 30:03
  SPEAKER_03: Wow.
• 30:04
  SPEAKER_03: Well, I mean I'm surprised as well to see it seems to be affecting like if you go down to the there's a graph there, Chris, and I think it might be interesting to see if they've got uh I think they've just got it by the models, the different models that are um kind of on the cost.
• 30:15
  SPEAKER_03: Yeah.
• 30:16
  SPEAKER_03: Um that one that one just above there.
• 30:17
  SPEAKER_03: Sorry, Chris, just went past.
• 30:18
  SPEAKER_03: Um, but what's interesting as well is that.
• 30:20
  SPEAKER_03: What the hell is that a graph?
• 30:21
  SPEAKER_03: That's a table.
• 30:22
  SPEAKER_03: Oh, sorry, sorry, table.
• 30:24
  SPEAKER_03: Um but it's it's the way that the the way that they've kind of got all of the, you know, you look at things like Mistral down the bottom um being 15 times, you know, and GPT-5 being 14 times.
• 30:34
  SPEAKER_03: And that most of the time that you're seeing at the top things like Gemini, Claude, you know, that it's it's in this sort of five to nine range.
• 30:41
  SPEAKER_03: Um, but it's just interesting to see some of those bigger models, even things like Mistral being quite, you know, uh ineffective.
• 30:47
  SPEAKER_03: Um but uh yeah, I think that's that's really interesting though.
• 30:50
  SPEAKER_03: I mean, I think there was something about um kind of the, I suppose the way that you articulate yourself as well in that language, it could be quite an interesting thing that even if we were to take out the like you're saying, take two sentences that uh or a sentence that can be compared, you know, directly.
• 31:02
  SPEAKER_03: But even if you just think about how people may might express themselves, um because somebody said to me about the about English as a language, uh they were a non-native English speaker, and they said, in one way it's great because it's uh there's there's fewer, you can use fewer words to say something that might take more words in say let's say French or or Spanish or whatever.
• 31:20
  SPEAKER_03: Um but that the problem is is that actually lots of native English speakers use too many words to communicate, you know.
• 31:26
  SPEAKER_03: Uh um Chris, we've talked about it when we write an email.
• 31:29
  SPEAKER_03: Uh uh German colleagues will say hi, then it's straight to what it would like you to do.
• 31:34
  SPEAKER_03: A British colleague will say, hi, good afternoon.
• 31:36
  SPEAKER_03: I hope you had a good weekend, hope this finds you well, blah, blah, blah, blah, blah.
• 31:39
  SPEAKER_03: All of that would be token usage.
• 31:41
  SPEAKER_03: This is kind of totally pointless.
• 31:42
  SPEAKER_03: So I just think it would be very, very useful to kind of um or interesting to see like the way that people communicate communicate in those languages, as well as just the sort of straight, you know, comparable token usage on a sentence.
• 31:55
  SPEAKER_03: Very interesting.
• 31:56
  SPEAKER_02: I think Ministrel is basically slower because it has a fac after every second question, and it basically just like, ah, I don't know.
• 32:03
  SPEAKER_02: French, it's a French model.
• 32:04
  SPEAKER_02: So basically, you it probably just waits for you to actually do things and has a copy in between, goes on strike, and uh sets fire on things.
• 32:11
  SPEAKER_02: Like these are uh I guess that's that's what's happening there.
• 32:15
  SPEAKER_02: I I've used Le Chat uh and especially the Shonk lately, uh, and it's actually a pretty pretty impressive tool.
• 32:21
  SPEAKER_02: Uh uh, but yeah, it's we don't talk about it too much uh enough, I think.
• 32:25
  SPEAKER_02: That's that's that's a bit annoying.
• 32:28
  SPEAKER_02: Um now uh the other thing is that I found interesting that there's not much actually uh payoff outside the tech sector.
• 32:37
  SPEAKER_02: I don't know who Apollo actually is.
• 32:39
  SPEAKER_02: Um I'm not surprised.
• 32:40
  SPEAKER_02: I'm not I'm not sure if that's actually a yeah, anyways, an economist that basically says like that uh in in IT, of course, AI is a big thing, but most of the other markets have haven't seen much profit changes because of it and because of the high expenditure that they do on AI.
• 32:56
  SPEAKER_02: Uh so Kate, you work in the construction environment with other companies there.
• 33:00
  SPEAKER_02: Do you see AI happening uh there and uh and being super successful?
• 33:04
  SPEAKER_02: Or is it something that the companies are still struggling to embrace?
• 33:08
  SPEAKER_01: Yeah, it's a very good question.
• 33:10
  SPEAKER_01: I would say that most of the most of the companies are still slightly afraid.
• 33:13
  SPEAKER_01: They're afraid of the data leaking, they're afraid of uh some possible like repercussions that could be caused by using AI.
• 33:20
  SPEAKER_01: And they are open to communicate with services that already have AI under the hood, like for example, our platform.
• 33:28
  SPEAKER_01: But in general, I would say that yeah, it's not very it's not a big thing there.
• 33:32
  SPEAKER_01: So, for example, if you're an architecture firm, how how exactly can you use AI?
• 33:36
  SPEAKER_01: Yes, you can plan something, but it's not like you can create the whole blueprint, you know, with it right now that will be approved.
• 33:44
  SPEAKER_01: So, like, yeah, it's not a big thing there.
• 33:48
  SPEAKER_02: Although it's interesting to see when you see these like like uh uh uh uh uh fast move uh uh 24-hour movies in China where people build a whole building in 24 hours, and like uh it seems like that a lot of that has been automated to to to the end's degree and that make them more effective.
• 34:05
  SPEAKER_02: I don't know.
• 34:07
  SPEAKER_01: Yeah, but it's not about AI, it's about automation, like proper automation, robots, and all of that stuff.
• 34:12
  SPEAKER_01: And I assume it has some AI under the hood, but still it's mainly like mechanical, I would say.
• 34:21
  SPEAKER_02: Well, you can also move faster if you don't care about people's safety or uh or work uh work uh work life balance or these kinds of things.
• 34:29
  SPEAKER_02: That's the other bit.
• 34:30
  SPEAKER_02: So it's always tricky to compare.
• 34:32
  SPEAKER_02: But yeah, it will be interesting to talk about that.
• 34:35
  SPEAKER_02: I'll be working with the German government in the uh in the next few weeks as well, uh, going through a meeting, a few meetups and talk about that.
• 34:41
  SPEAKER_02: So I'm gonna take a look at like how that works for government in uh initiatives.
• 34:45
  SPEAKER_02: I wouldn't be surprised if they also realized we spend a lot of money on AI, but we don't get much back.
• 34:52
  SPEAKER_05: Yeah.
• 34:54
  SPEAKER_02: Uh a few geek things to talk to, and I'm gonna go through them quickly.
• 34:58
  SPEAKER_02: This is actually really cute.
• 34:59
  SPEAKER_02: Uh local Wikipedia, where you basically uh find uh only Wikipedia entries 1.5 miles around you where you are.
• 35:08
  SPEAKER_02: And I remember winning a hackathon in 2009 uh uh with the same thing because Wikipedia always had an option to search in uh if in a locale, and uh that's something that a lot of people uh don't realize, but this is actually really cute to just see where you are and what's happening around you.
• 35:25
  SPEAKER_02: So that's kind of cute to get like your own Wikipedia for where you are in the moment.
• 35:29
  SPEAKER_02: So that's pretty cute.
• 35:32
  SPEAKER_02: In 2008, I was giving a training in uh in California about uh Yahoo News and the Yahoo News API, and that's where we use that Wikipedia stuff as well.
• 35:42
  SPEAKER_02: And I just showed people how to do an RSS reader and basically should display an RSS feed, and there was this uh uh this headline that Russia invaded Georgia because it was 2008 when they invaded Georgia, and uh then people five people in the audience were all like, oh my god, that's horrible.
• 35:58
  SPEAKER_02: And one of them, like, yeah, my my aunt lives in Atlanta, and and I'm like, no, not that Georgia.
• 36:10
  SPEAKER_00: Yeah.
• 36:11
  SPEAKER_03: I just like the fact their reaction was, yay, my auntie lives in in Georgia.
• 36:15
  SPEAKER_03: It's like they really don't like their auntie then, do they?
• 36:17
  SPEAKER_03: That's amazing.
• 36:18
  SPEAKER_02: I'm also I'm also more confused.
• 36:20
  SPEAKER_02: Like, how would you think that another country would invade one of your states and not like your whole country?
• 36:26
  SPEAKER_02: That's the other thing that did my head in like when in this person's reaction.
• 36:29
  SPEAKER_02: I'm like, no, it's not like Russia is gonna invade only Bavaria and Germany or uh or or like you know, or like only goes to Wales and not England.
• 36:38
  SPEAKER_02: Like, you know, like I'm not sure.
• 36:41
  SPEAKER_02: Really bizarre.
• 36:44
  SPEAKER_02: Uh you can query large language model over DNS.
• 36:49
  SPEAKER_02: Because why not?
• 36:51
  SPEAKER_02: Yeah, that's exactly what I thought.
• 36:53
  SPEAKER_03: It's just developers.
• 36:54
  SPEAKER_03: I mean, Kate, Kate, I I know you'll have seen previous shows, but we we absolutely love these projects where you're not entirely sure why they exist or why somebody's made them, but you're sort of glad they have, right?
• 37:04
  SPEAKER_03: And I think that that is a really good example of it.
• 37:05
  SPEAKER_03: It's just I what I love is this is the epitome of a I wonder if you could type of project.
• 37:13
  SPEAKER_03: Do you know what I mean?
• 37:14
  SPEAKER_03: And that's what I love is the like this is weird, and and I always go back to bursting the bubble that we're in of when we build tools like this that we know other developers are gonna love, and then you have to explain them to somebody that's just seeing you doing your job and they're going, What are you working on?
• 37:28
  SPEAKER_03: And it's like you you can't you can't really explain it, can you?
• 37:30
  SPEAKER_03: So, you know, it's our it's our it's our little thing that we enjoy, and that's it.
• 37:34
  SPEAKER_03: It doesn't make sense to anyone else.
• 37:36
  SPEAKER_01: No, of course, yeah, of course, and that's why it also makes me sad that Microsoft closed the garage because that was a place for experimentations like that, you know.
• 37:44
  SPEAKER_01: When you have a Ryan production uh application, you don't have time to experiment, and when you go home, you also don't have time to do that.
• 37:51
  SPEAKER_01: So, yeah, that's really cool that the things exist.
• 37:54
  SPEAKER_01: I agree.
• 37:55
  SPEAKER_03: So great.
• 37:56
  SPEAKER_03: It's got such a clawed look, though, doesn't it?
• 37:58
  SPEAKER_03: I did that's the one thing I kind of go, uh that would be the first thing I'd be prompting away is the uh the the clawed look, you know, that that site had because it just looks like it's part of the kind of anthropic, even down to like the colours and the fonts.
• 38:09
  SPEAKER_03: Look, that just is straight up um anthropic.
• 38:11
  SPEAKER_03: Um so you know, maybe that was the point.
• 38:13
  SPEAKER_03: But also one more thing, just to jump on this.
• 38:15
  SPEAKER_03: These blocks, you know, this feels like it's going back to the kind of um, like you said, Chris before, the kind of big hero image at the top, three blocks underneath.
• 38:22
  SPEAKER_03: These are you know our services, like the old layouts that we used to have of the there, just that that those um those kind of all there again.
• 38:29
  SPEAKER_03: I'm seeing a lot of this of like features being uh laid out on on landing pages with those, with just like these this grid kind of block.
• 38:37
  SPEAKER_03: Um it just seems to me it's become more of a trend again.
• 38:39
  SPEAKER_03: I can't quite put my finger on it.
• 38:41
  SPEAKER_03: That why the maybe it is just the uh clawed default, you know, that it puts things into these these tidy little boxes.
• 38:47
  SPEAKER_02: But yeah, I'm also confused why it has an overflow and why it actually doesn't do flex box and just show them under each other.
• 38:54
  SPEAKER_02: I think that seems to be a bug to me, but okay.
• 38:59
  SPEAKER_01: I think it might also be related to, for example, V0 from Versal, because they also implement a lot of designs that look like this, you know, like exactly the same blocks with some features under the hood.
• 39:10
  SPEAKER_01: So it might be also yeah, developed there.
• 39:14
  SPEAKER_02: It's a bit like when you had the old the the original WordPress template, like the Kubrick one with the blue rectangle on the top, and then basically everything underneath, and for decades every block looks the same, so that makes sense as well.
• 39:27
  SPEAKER_02: This is really cool.
• 39:28
  SPEAKER_02: There's a there's a new frame, uh, new social network called Def Null, and that's only for developers.
• 39:35
  SPEAKER_02: And I did sign up for it, and it actually is threaded and it's geeky as hell.
• 39:40
  SPEAKER_02: And uh I find it very fascinating because it's only the people that I have in my inbox every day, anyways, that I meet there at the moment, but seems like people are starting to get into that, so it's uh it's a pretty cool.
• 39:52
  SPEAKER_02: Uh you have to sign in with GitHub, and you can see basically all your uh all your posts are pull requests, and people then can uh uh can change them as well.
• 40:03
  SPEAKER_02: And it's all threaded and it's quite fun.
• 40:06
  SPEAKER_03: That's an interesting idea.
• 40:08
  SPEAKER_03: That's a really cute pull request.
• 40:10
  SPEAKER_03: Yeah.
• 40:10
  SPEAKER_03: Is it using so is it using uh the sort of GitHub it because I know you can do that with like uh you know I've done it on my blog where it's like uh bringing in um the you can use your GitHub comments on on the kind of repo as your comments on the blog.
• 40:23
  SPEAKER_03: So it just kind of uses that that infrastructure.
• 40:25
  SPEAKER_03: I wonder if that's maybe what they're doing, maybe that's why the pull requests like is it being served from GitHub or you know being stored there?
• 40:32
  SPEAKER_03: Just be interested to know.
• 40:33
  SPEAKER_02: Yeah, and I I I think it's also using the add proto under the hood, uh uh and all kind of that cool Fediverse thing that is weird.
• 40:42
  SPEAKER_02: Um cool.
• 40:44
  SPEAKER_02: Um tips and tricks.
• 40:45
  SPEAKER_02: I didn't have time to finish the video, just two things that are happening right now that actually are interesting.
• 40:51
  SPEAKER_02: I love that uh Chris uh uh Chris um from Master Dev or former CSS tricks did an analysis on like the social media images that you auto-generated, what you should show up in them, because that's something that I do right now with redesigning my blog.
• 41:08
  SPEAKER_02: And uh he also shows you like which platforms do show what and which platforms don't show anything.
• 41:16
  SPEAKER_02: So this is a really cool little thing to realize like why people don't have an automatic preview of your stuff, and there's so many things that are there, and WhatsApp.
• 41:26
  SPEAKER_02: I didn't even know the WhatsApp does that as well.
• 41:28
  SPEAKER_02: So um, yeah, it's pretty pretty amazing and explains the unfurling, what that means, and traffic-wise, and how you should do the images.
• 41:36
  SPEAKER_02: And it's a good thing to uh to if you want to rebrand your product to actually be successful.
• 41:41
  SPEAKER_02: Um yeah, it's a good opportunity to to get that done.
• 41:45
  SPEAKER_02: And there's something we we've done for our magazine as well, but we haven't done for all the other bits.
• 41:50
  SPEAKER_02: And uh, Dan, you're working on a lot of like image generators right now for VR developers, so that's something to get in there as well.
• 41:58
  SPEAKER_02: And the other thing is that CSS random is now supported everywhere and not only in Safari.
• 42:03
  SPEAKER_02: So you can do random stuff right now in CSS, which is pretty amazing because randomness is fun.
• 42:10
  SPEAKER_02: Uh, it's never I mean the fun thing is when you do anything with random, you can count on your fingers the seconds down until the first developer will come up, but it's not really random, is it?
• 42:22
  SPEAKER_05: Yeah.
• 42:26
  SPEAKER_03: It was gonna be my first question as well.
• 42:27
  SPEAKER_03: You know, when it said like random is that I was gonna go, is it really random though?
• 42:30
  SPEAKER_03: You know, it is it's so predictable, isn't it?
• 42:32
  SPEAKER_03: It's like we're pedantic, but it isn't random though.
• 42:35
  SPEAKER_01: So but actually, what is random, guys?
• 42:38
  SPEAKER_01: Like in general, you know, when you toss a coin, is it really random?
• 42:42
  SPEAKER_01: No, it's chaos, actually.
• 42:44
  SPEAKER_01: It's not random.
• 42:45
  SPEAKER_01: You can, if you know all of the circumstances, all of the conditions, you can predict it.
• 42:50
  SPEAKER_01: So, like, what do we call random, you know?
• 42:55
  SPEAKER_03: That's true.
• 42:55
  SPEAKER_03: You have a better argument than I did there, so yeah, I concede.
• 42:58
  SPEAKER_01: But actually, yeah.
• 43:01
  SPEAKER_01: Yeah, sorry, this is really cool about this random thing.
• 43:03
  SPEAKER_01: And I think it's fantastic that CSS implements so many functions recently.
• 43:08
  SPEAKER_01: I saw so many news about that, that CSS is going forward with lots of functions, and I think it's fantastic, and it will definitely improve lots of performances in lots of performance in lots of websites if people switch from I don't know, like tailwind or material UI to um uh claims CSS.
• 43:28
  SPEAKER_03: Yeah, it's true.
• 43:29
  SPEAKER_03: I mean, I think also it does make a good point because he he says right at the start of the article, something along the lines of like, um, random is now a feature, but uh I had to sort of ask myself where where would I use it?
• 43:38
  SPEAKER_03: Like what are the use cases?
• 43:39
  SPEAKER_03: Um and I think uh Chris, I straight away went to kind of we talked about this quite a lot that it's like in it's in our nature to want to be uh even the phrase pixel perfect, like it's it's not really a positive thing because it's just saying, I get to decide how this thing looks for you.
• 43:53
  SPEAKER_03: And I think that this this randomness at least will kind of make us think a little bit differently.
• 43:58
  SPEAKER_03: Um not about everything.
• 44:00
  SPEAKER_03: But you know whether or not we need something to be controlled or just to a degree random.
• 44:05
  SPEAKER_02: Yeah, or like freshness as well.
• 44:07
  SPEAKER_02: I mean, using CSS, I mean, not in CSS, of course, but why why not in CSS?
• 44:12
  SPEAKER_02: You can actually change the order of things now randomly without having to do a reload.
• 44:18
  SPEAKER_02: And that would be one thing to use it for, kind of thing.
• 44:21
  SPEAKER_02: And anything that has to do with like simulating nature had to do some random function, like any particle system, anything when you did uh when you did smoke, when you actually do sparkles, these kind of things all need random.
• 44:35
  SPEAKER_02: And uh, I mean it was never really that much of an issue because you can actually uh use a CSS uh uh variable and access that one in JavaScript and just get random from JavaScript and put it into that one.
• 44:46
  SPEAKER_02: Uh, but now we have it natively, and I think it's gonna be interesting.
• 44:50
  SPEAKER_02: And uh it's funny because real developers have always poo-pooed on C CSS of being a toy language, and now it's becoming so complex that they're that they say CSS is too hard to learn.
• 45:02
  SPEAKER_02: So make up your mind.
• 45:03
  SPEAKER_02: Like it's really tricky to actually embrace it or not, but uh it's fun that you can specialize in being a CSS developer nowadays, and and it's not seen as something like that you're being lazy, but there's so many things in CSS to not to start uh understanding.
• 45:19
  SPEAKER_02: Cool.
• 45:20
  SPEAKER_02: Talking about random, let's start playing fake on news, uh our game where we actually give you headlines, and you have to tell us if these headlines are real or if I made them up, if they're random.
• 45:31
  SPEAKER_02: Uh, and some of the real ones might be random, so you actually have an issue there.
• 45:35
  SPEAKER_02: Uh, it's got so there's also gonna be uh uh there's there's there's there's a list of people who played it before.
• 45:41
  SPEAKER_02: You can win absolutely nothing, but you're gonna have the bragging rights of being number one if you actually get to number one.
• 45:47
  SPEAKER_02: So let's play fake on news.
• 45:49
  unknown: I love it.
• 46:01
  SPEAKER_03: All right.
• 46:02
  SPEAKER_03: Let's see if that one works for the first time.
• 46:05
  SPEAKER_03: Fingers crossed.
• 46:06
  SPEAKER_03: Okay, so you have 30 seconds for each headline, and Chris will read it out, and then I'll help you uh come to a conclusion that the the final choice is of course yours.
• 46:15
  SPEAKER_03: All right, let's play.
• 46:20
  SPEAKER_02: Can you say that?
• 46:21
  SPEAKER_00: No, that's fake.
• 46:23
  SPEAKER_02: Grand Theft AI.
• 46:24
  SPEAKER_02: Highway thieves are stealing millions in data center gear.
• 46:28
  SPEAKER_02: I think it's fake.
• 46:30
  SPEAKER_01: Yeah, I think it's fake.
• 46:31
  SPEAKER_03: It feels quite real though, data centers all over the US.
• 46:35
  SPEAKER_03: You're not convinced?
• 46:37
  SPEAKER_00: No, I think it's fake.
• 46:38
  SPEAKER_03: Okay, conviction, let's see.
• 46:41
  SPEAKER_03: It's actually real.
• 46:43
  SPEAKER_03: Sorry.
• 46:45
  SPEAKER_00: Okay.
• 46:46
  SPEAKER_02: Okay, onwards and upwards.
• 46:48
  SPEAKER_02: It's offensively cheap how solar power is quietly becoming bigger than nuclear.
• 46:56
  SPEAKER_01: No, that's fake.
• 46:58
  SPEAKER_01: It's not true.
• 46:59
  SPEAKER_01: It's not becoming bigger than nuclear.
• 47:02
  SPEAKER_03: Oh, see, I'm gonna disagree with you, but I I'll of course I'll let you let you choose.
• 47:07
  SPEAKER_03: But I I've got a feeling, I don't know, solar feels like it's the time that maybe it's actually real.
• 47:14
  unknown: Oh my god!
• 47:16
  SPEAKER_02: The problem is the more solar panels we install, the the weaker the sun will become.
• 47:21
  SPEAKER_02: So that's the issue that we have with that, but yeah.
• 47:24
  SPEAKER_02: Okay, number three.
• 47:28
  SPEAKER_02: Open clock creator says he spent 10 million in tokens last month.
• 47:37
  SPEAKER_03: I mean, really?
• 47:38
  SPEAKER_03: That feels why would he spend 10 million?
• 47:39
  SPEAKER_03: I suppose is he is he paying for everybody?
• 47:42
  SPEAKER_03: I mean, how does that even work?
• 47:43
  SPEAKER_03: I don't know.
• 47:43
  SPEAKER_03: What do you feel on that, Kate?
• 47:45
  SPEAKER_01: I would say that this is too this is fake.
• 47:48
  SPEAKER_03: Yeah, it feels really 10 million, that'd be a hundred million a year as well.
• 47:52
  SPEAKER_02: Surely not it's actually real.
• 47:56
  SPEAKER_02: Well, he says it, if it's true, we don't know.
• 47:58
  SPEAKER_02: But like Wow.
• 48:01
  SPEAKER_03: Okay, well, we'll we'll we have we yet to pick up a point, but I think this is gonna be the one, number four.
• 48:07
  SPEAKER_02: Following the PHP to shell security incident, WordPress switches to Python in next release.
• 48:15
  SPEAKER_01: No, no, it's fake.
• 48:18
  SPEAKER_01: It's it cannot be too so fast to to switch to Python.
• 48:23
  SPEAKER_01: I would say it's fake.
• 48:25
  SPEAKER_03: Agreed, isn't it?
• 48:25
  SPEAKER_03: Something like I don't know, 40% of the internet or something is WordPress and PHP.
• 48:29
  SPEAKER_03: It can't be.
• 48:31
  SPEAKER_03: It is fake.
• 48:38
  SPEAKER_03: Let's keep it going.
• 48:39
  SPEAKER_03: Number five.
• 48:40
  unknown: Yeah.
• 48:41
  SPEAKER_02: Meta smart glasses were recording conversations without the wake phrase company admits.
• 48:47
  SPEAKER_01: I can't believe that.
• 48:48
  SPEAKER_01: I would say that it might be true.
• 48:50
  SPEAKER_03: Yeah, I feel like it is as well.
• 48:52
  SPEAKER_03: I'm just hoping that he's not luring us into a trap.
• 48:56
  SPEAKER_03: It is true.
• 48:57
  SPEAKER_01: It is true.
• 48:58
  SPEAKER_03: Alright.
• 49:00
  SPEAKER_02: Number six.
• 49:02
  SPEAKER_02: Shiam Chanka, CTO of Palantir, says AI researchers have started smoking because they're convinced AI, Claude, probably is going to cure cancer.
• 49:13
  SPEAKER_03: Surely not.
• 49:15
  SPEAKER_00: No, I would say it's fake.
• 49:19
  SPEAKER_03: I'm just hoping, you know, developers are intelligent people.
• 49:22
  SPEAKER_03: Surely they wouldn't be doing this.
• 49:24
  SPEAKER_03: Let's see.
• 49:25
  unknown: Oh my god.
• 49:26
  SPEAKER_03: It's actually news.
• 49:29
  SPEAKER_01: No way.
• 49:30
  SPEAKER_01: No way.
• 49:31
  SPEAKER_02: Again, he said it.
• 49:32
  SPEAKER_02: If it's true, we don't know.
• 49:33
  SPEAKER_02: But like he said it.
• 49:35
  SPEAKER_01: Okay.
• 49:36
  SPEAKER_01: It's true that she said.
• 49:37
  SPEAKER_01: Okay, okay, okay.
• 49:39
  SPEAKER_03: Yeah, you gotta look out for the technicalities.
• 49:41
  SPEAKER_03: Alright, number seven.
• 49:43
  SPEAKER_02: Apple releases CSS library to make it easier for web designers to support the iPhone Duo.
• 49:49
  SPEAKER_02: I've not heard about it.
• 49:51
  SPEAKER_01: No, I think it's fake.
• 49:53
  SPEAKER_01: It's also too fast.
• 49:55
  SPEAKER_03: Yeah.
• 49:56
  SPEAKER_03: On a CSS library?
• 49:57
  SPEAKER_03: I don't I don't know.
• 49:58
  SPEAKER_03: No, I don't think so.
• 50:00
  SPEAKER_02: It is fake.
• 50:02
  SPEAKER_02: Could be good though.
• 50:04
  SPEAKER_02: Number eight.
• 50:06
  SPEAKER_02: Grey paint has never had as good as in the AI boom.
• 50:12
  SPEAKER_03: I really don't know what that even means.
• 50:14
  SPEAKER_03: Why why would that be the case?
• 50:16
  SPEAKER_01: No, I would also say it's fake.
• 50:20
  SPEAKER_03: Although, although when I I'm I'm wanting to go with news for that reason, because it feels a bit like it could kind of slip through the net.
• 50:28
  SPEAKER_03: What do you think?
• 50:30
  SPEAKER_00: Fake.
• 50:32
  SPEAKER_00: I'm saying it's fake.
• 50:34
  SPEAKER_02: It's actually news.
• 50:35
  SPEAKER_02: There's a special gray paint that is fireproof and actually very optimized for data centers.
• 50:41
  SPEAKER_02: That's why everybody is actually ordering that paint from that one provider in America.
• 50:45
  SPEAKER_02: You could have seen that it's actually news because it's actually written not in British English.
• 50:50
  SPEAKER_02: So that's something that I don't write.
• 50:52
  SPEAKER_03: Yeah, the gray.
• 50:54
  SPEAKER_03: But also I did think the a dot i a dot i dot was a bit you know unusual, but there you go.
• 50:59
  SPEAKER_03: Number nine.
• 51:01
  SPEAKER_02: Warhound raises 200 million dollars to engineer war dogs and arm them with Andoril gear.
• 51:09
  SPEAKER_03: I mean it feels real with a name like Warhound.
• 51:13
  SPEAKER_01: I would say it could be real.
• 51:15
  SPEAKER_01: Let's say it's real.
• 51:18
  SPEAKER_03: Yeah, it just sounds so bonkers that it does feel like it's real.
• 51:22
  SPEAKER_02: Yeah, we're getting there.
• 51:24
  SPEAKER_02: I didn't I didn't understand a single word of that headline, but it's real.
• 51:27
  SPEAKER_02: It's like what is Andrew gear?
• 51:29
  SPEAKER_02: I have no idea, but okay, fair enough.
• 51:32
  SPEAKER_02: Yeah.
• 51:32
  SPEAKER_02: Now I'm better.
• 51:34
  SPEAKER_02: Apple will reportedly run diagnostics on returned iPhones without opening the packaging.
• 51:42
  SPEAKER_03: That feels like it could be real.
• 51:43
  SPEAKER_03: I mean, perhaps.
• 51:45
  SPEAKER_03: Using yeah, scanning of some kind.
• 51:47
  SPEAKER_01: How can you do that?
• 51:48
  SPEAKER_01: But okay, let's say it's real.
• 51:49
  SPEAKER_01: Yeah, it might be the case.
• 51:51
  SPEAKER_01: It's Apple.
• 51:52
  SPEAKER_03: Yeah.
• 51:55
  unknown: Okay.
• 51:55
  SPEAKER_01: It's real.
• 51:57
  SPEAKER_01: Okay.
• 51:57
  SPEAKER_02: I've seen also people walk around in San Francisco and do like Bluetooth scanning to see if you have uh a Macbook, a MacBook in your boot of your car.
• 52:06
  SPEAKER_02: So uh that's that's another thing to get that done.
• 52:08
  SPEAKER_02: So basically to realize that if some some gear is in the car to make it worthwhile, breaking it open.
• 52:13
  SPEAKER_05: Wow.
• 52:14
  SPEAKER_05: Okay.
• 52:15
  SPEAKER_02: Number 11.
• 52:17
  SPEAKER_02: Concerned about AI security, German enterprise customers resort to USB sticks for secure updates.
• 52:25
  SPEAKER_03: I could see that.
• 52:26
  SPEAKER_03: I could see that happening.
• 52:27
  SPEAKER_01: Ah yeah, yeah, I could see that.
• 52:29
  SPEAKER_01: I used to work with German customers.
• 52:31
  SPEAKER_01: I would say it's it's true.
• 52:33
  SPEAKER_03: Yeah, I also think I I work with some companies like you do, Kate, that are construction, and they they are still using USBs, and it's some of it's oh great, okay, okay.
• 52:43
  SPEAKER_03: No, you load us in there.
• 52:45
  SPEAKER_02: You load us in.
• 52:46
  SPEAKER_02: Well, you US nuclear power until the last until two years ago used floppy disks for that because of that reason, and uh data changed around as well.
• 52:56
  SPEAKER_02: But so it's not that uncommon.
• 52:58
  unknown: Okay.
• 52:58
  SPEAKER_02: AI agents use idle time to index root folders, researchers say.
• 53:09
  SPEAKER_01: I would say it's fake.
• 53:11
  SPEAKER_01: Why would you do that?
• 53:13
  SPEAKER_03: Yeah, surely not.
• 53:17
  SPEAKER_02: It is fake, luckily enough.
• 53:19
  SPEAKER_02: Cool.
• 53:20
  SPEAKER_02: Well, let's see how you fair.
• 53:22
  SPEAKER_02: Let's see.
• 53:24
  SPEAKER_03: Oh, seventh.
• 53:25
  SPEAKER_03: Oh, seventh! Okay, not bad.
• 53:28
  SPEAKER_03: Very good.
• 53:28
  SPEAKER_03: Yeah, yeah.
• 53:30
  SPEAKER_01: I'm fine with losing.
• 53:32
  SPEAKER_01: Yeah, it's not a problem.
• 53:33
  SPEAKER_01: It was very interesting, though.
• 53:34
  SPEAKER_01: I didn't expect me to be so excited about that.
• 53:39
  SPEAKER_02: Cool.
• 53:41
  SPEAKER_02: Well, that was fake good news for today.
• 53:43
  SPEAKER_02: Uh, so now we're gonna go into your topic.
• 53:45
  SPEAKER_02: And one thing that I talked to Dan about uh when it comes to biases, um let me share my screen or stop sharing yours, or what do we do here?
• 53:56
  SPEAKER_02: Um hang on.
• 53:57
  SPEAKER_02: Oh yeah, yeah.
• 53:59
  SPEAKER_02: Uh this is a classic case about biases, and this is basically in in the second world war, uh they noted down uh bullet holes in aircraft that came back from combat.
• 54:11
  SPEAKER_02: So uh to see how to actually make the the aircraft more secure.
• 54:16
  SPEAKER_02: So these were the bullet holes that came in, and the biggest thing that basically were the bias was where people said, like, okay, where the bullet holes are, this is where we're gonna uh that's where we're gonna fortify the planes to make them secure and make sure that they can't be shut down.
• 54:32
  SPEAKER_02: But that was the bias of like having the information and having recorded that information and knowing that information.
• 54:39
  SPEAKER_02: Because what they forgot about is like the planes that came back where they could note down the bullet holes, were the ones that weren't shut down.
• 54:47
  SPEAKER_02: So basically, where you should fortify that plane was not actually where the bullet holes were, but where they weren't.
• 54:56
  SPEAKER_01: But that was like it's a classic case, yeah.
• 54:59
  SPEAKER_01: It's a classic survivorship bias.
• 55:00
  SPEAKER_01: And I to be honest, I see it so often, yeah, in in among developers, for example, we try to follow some kind of um framework or any kind of library that we see only good news about, and we don't know anything about companies that didn't succeed, for example, with this specific technology, and we follow some who succeed, and we don't read anything about those who didn't.
• 55:24
  SPEAKER_01: So, yeah.
• 55:26
  SPEAKER_02: So survivorship bias is the name of that one.
• 55:28
  SPEAKER_02: And if you now ask an A uh I I would assume that if you ask an LLM to actually say, like, where should we fortify the plane or what should we do to the plane to actually make it secure, it would probably come up with something like that.
• 55:45
  SPEAKER_02: That is brilliant.
• 55:47
  SPEAKER_03: That's very good.
• 55:49
  SPEAKER_03: But yeah, let's let's talk about No, I was just gonna, I was just gonna I was gonna sort of jump on that as well, really, and just sort of saying, so so Kate, when you kind of, because you delivered this this session at the World Congress uh North America, didn't you?
• 56:00
  SPEAKER_03: And um, you know, I know we kind of had a brief chat about it, but what was the what was the the the point when you decided that was the topic you wanted to to kind of um you know deep dive into and do a session on?
• 56:09
  SPEAKER_03: What was it a bias that you saw in yourself, other people, or just a topic you've always been interested in?
• 56:14
  SPEAKER_01: Yeah, like I'm in general very curious about how our brains work, in general, like as humans, and I see like so many issues with how our brains work just by evolution uh wired.
• 56:26
  SPEAKER_01: And um, there are more than 170 cognitive biases, actually, and they already, most of them are well studied.
• 56:33
  SPEAKER_01: And one of the most popular, I would say, is obviously the confirmation bias, and the second one is some cost fallacy.
• 56:39
  SPEAKER_01: And we always, always as developers, we face them.
• 56:42
  SPEAKER_01: And I see it all the time among myself and among other developers that I work with.
• 56:47
  SPEAKER_01: And the sun cost fallacy is one of my favorites just because I see how much time we could spend doing something that goes nowhere, and how much how many resources we could spend doing something that is not needed.
• 57:01
  SPEAKER_01: So, and that time, the problem is that the time that we spend, it feels good, it feels useful, it feels that we are doing something important, but we we don't.
• 57:12
  SPEAKER_01: We don't do anything important when we follow the SANCOS fallacy.
• 57:16
  SPEAKER_01: So, yeah, I just really wanted to share this and help other people also to recognize those among themselves.
• 57:23
  SPEAKER_02: I mean, it also goes, uh I mean, the sound cost thing also goes where people stay longer in jobs than they shouldn't.
• 57:29
  SPEAKER_02: I mean, so many times I've seen developers being being continuously disappointed by their company by not allowing them to do the things that they want to do, but they keep staying for like, oh, next month it's gonna be that they're gonna allow me to do the thing properly because I've I've been here for eight years, so they probably listen to me because of the eight years.
• 57:48
  SPEAKER_02: But it's again like you put the more effort you put in, the more comes out of it.
• 57:52
  SPEAKER_02: It doesn't mean that it's gonna be changing anytime soon.
• 57:55
  SPEAKER_02: More likely it's actually not.
• 57:56
  SPEAKER_02: So um yeah, that's the that's a really interesting one.
• 58:00
  SPEAKER_02: As I said, like there's when you when you go into neuroscience and also you go into people that work in social environments, they also know all know about biases.
• 58:08
  SPEAKER_02: And uh when you do like uh trainings on like uh leadership trainings as well.
• 58:12
  SPEAKER_02: We in Microsoft there was a lot of stuff, in Yahoo, there was a lot of stuff, Mozilla as well.
• 58:17
  SPEAKER_02: Um, so what are the the main thing, the main ones that you see happening, and how can somebody prevent himself from falling into that trap to have that bias?
• 58:28
  SPEAKER_01: Uh yeah, I would say, as I mentioned, the confirmation bias is the must one, the most um important, I would say, for developers to know.
• 58:37
  SPEAKER_01: And the second one is anchoring bias.
• 58:39
  SPEAKER_01: So this one is also very interesting.
• 58:41
  SPEAKER_01: What is happening is that when we hear something and when we see some kind of number, it sticks with us.
• 58:47
  SPEAKER_01: And then the next prediction we make, we unconsciously make based on that number.
• 58:53
  SPEAKER_01: For example, someone on a plane and just casually throws something like, Yeah, this feature will take two days.
• 59:01
  SPEAKER_01: And then the next developer will be uh considering these two days in their job, you know, without even diving into the feature, without even knowing about that.
• 59:11
  SPEAKER_01: They just, ah yeah, two days, and they they don't even know why they think two days, or like three, something around that.
• 59:17
  SPEAKER_01: And there were lots of research, uh lot of research about that.
• 59:20
  SPEAKER_01: And unfortunately, as people, we cannot eliminate those biases in us.
• 59:25
  SPEAKER_01: The only thing we can do is make us aware that they exist and how we can work around.
• 59:32
  SPEAKER_01: And the main uh, I would say, uh the main tool that we can use is awareness, of course.
• 59:37
  SPEAKER_01: We should be aware of what is happening in our brain, how our thoughts are being dealt, why do I think what I think, and what do I know what I know?
• 59:48
  SPEAKER_01: Uh, and this is one of the most important.
• 59:50
  SPEAKER_01: And the second one is I think that we should all be as a scientist, we should all behave as scientists.
• 59:55
  SPEAKER_01: So we should we should always ask ourselves what could um what kind of assumption can I have that would break my belief?
• 1:00:05
  SPEAKER_01: So I think that this is a front end back.
• 1:00:08
  SPEAKER_01: What kind of log can I find that would that would tell me that no, it's not front end, it's backend, for example, or something else.
• 1:00:16
  SPEAKER_01: So what could break my belief?
• 1:00:18
  SPEAKER_01: And this is what actually scientists do, and this is what everyone should do, not only developers actually, but for developers, I think it's actually critical.
• 1:00:26
  unknown: I think this is.
• 1:01:54
  SPEAKER_01: Yeah, it's actually a really good question.
• 1:01:56
  SPEAKER_01: I was also thinking about if AI has our biases in general, you know.
• 1:02:02
  SPEAKER_01: But it it works a bit different with AI because AI doesn't have, I would say, like consciousness in in the way that we have as humans, right?
• 1:02:09
  SPEAKER_01: And it doesn't have the same um like wired connections in it's a let's say brain, right?
• 1:02:15
  SPEAKER_01: So we it's basically it's basically it's being trained on the data that we had.
• 1:02:19
  SPEAKER_01: And I think it's not more critical, it's critical in the same amount as it was before, like to be aware of what is happening in our brain.
• 1:02:28
  SPEAKER_01: So uh we like as AI, AI is a tool, and we need to study, let's say, how to use it, because right now AI is everywhere, and everyone became a specialist of how to use AI, how to write prompts, how to do all of that stuff.
• 1:02:42
  SPEAKER_01: And I don't think it's the case.
• 1:02:44
  SPEAKER_01: Like lots of people have no idea how to use AI, even even in uh even in even among developers.
• 1:02:49
  SPEAKER_01: I see it on a day-to-day basis how people use it, and I'm like, guys, what are you doing?
• 1:02:53
  SPEAKER_01: Like it's maybe I'm wrong, maybe you're like, what is happening here?
• 1:02:58
  SPEAKER_01: And yeah, what we need to make sure is just we should check everything that is critical.
• 1:03:03
  SPEAKER_01: We should check every assumption that we have uh that AI gives us.
• 1:03:10
  SPEAKER_02: I think the the uh one of the things is kind of uh the when you work with people who work in the social field, the first thing, the first two years in their education, what they do is actually ask themselves and question themselves and question their biases.
• 1:03:23
  SPEAKER_02: Oftentimes we don't want to be seen as a biased person, so we actually don't face or don't question why we made a certain decision.
• 1:03:33
  SPEAKER_02: How do we break that barrier?
• 1:03:34
  SPEAKER_02: How do we break that kind of like feeling uh feeling embarrassed for having found out that we have a bias?
• 1:03:42
  SPEAKER_01: But the thing is that I think it's about accepting that I am a human, I have biases.
• 1:03:49
  SPEAKER_01: This is what in my is in my nature.
• 1:03:53
  SPEAKER_01: It's not possible to not have biases, and just being afraid of having them means that you as a person have no idea how you as a human work, you know.
• 1:04:04
  SPEAKER_01: So it's not possible to not have them.
• 1:04:06
  SPEAKER_01: Everyone has them.
• 1:04:08
  SPEAKER_01: And as with mistakes, we should be open to make those mistakes with AI, except including with AI.
• 1:04:16
  SPEAKER_01: Like, I mean, okay, I have biases, that's fine.
• 1:04:19
  SPEAKER_01: I don't like it, I don't want to have those, but I will do my best to understand myself.
• 1:04:25
  SPEAKER_01: And what's also very interesting is that there is a specific bias that when you learn biases, when you learn about them, you start to notice those in other people.
• 1:04:36
  SPEAKER_01: And this is one of the most interesting moments I find because I also, when I started to study those, I was like, okay, so this friend of mine has a confirmation bias right now, and this friend of mine has this one.
• 1:04:47
  SPEAKER_01: And I became such an annoying person at this point because I was so arrogant, thinking that now I know everything, you know.
• 1:04:54
  SPEAKER_01: And it was also a moment for me to be more humble and accepting that, yeah, it's just because I know that biases exist and I have them too.
• 1:05:03
  SPEAKER_01: You know, so yeah, like it's just awareness and understanding that we all have them and it's fine, and we should we should I don't know, like be more curious about ourselves, and we should kind we should be kind to ourselves because it's so hard to um not be judgmental to other people and to ourselves as well.
• 1:05:21
  SPEAKER_01: So I'm fine, I have biases, I am a person.
• 1:05:24
  SPEAKER_01: That's completely fine.
• 1:05:25
  SPEAKER_01: You too, everyone has, and we should be open about that.
• 1:05:30
  SPEAKER_02: You too has biases?
• 1:05:31
  SPEAKER_02: I'm a bonal notes.
• 1:05:37
  SPEAKER_03: You'll never get a pun pass, Chris.
• 1:05:38
  SPEAKER_03: You'll never get a pun pass from the past.
• 1:05:43
  SPEAKER_02: But yeah, I mean you're totally you're totally right.
• 1:05:45
  SPEAKER_02: The problem with LLMs, especially the free ones, are sometimes two years, uh two to three years in the past.
• 1:05:50
  SPEAKER_02: So they've been trained.
• 1:05:52
  SPEAKER_02: Uh the the the training data that went in, that one led to a lot of biases.
• 1:05:56
  SPEAKER_02: I remember even in hardware, when for example, like doors would.
• 1:06:00
  SPEAKER_02: Open for people with dark skin because they only trained for facial recognition with people of white skin.
• 1:06:06
  SPEAKER_02: So that's that those kind of things happen.
• 1:06:08
  SPEAKER_02: And so you don't want to build racist software, you don't want to build uh biased software by just basically assuming that the data that went in was diverse enough.
• 1:06:19
  SPEAKER_02: So that's a big thing that I see a lot of people that we we actually have racist AI out there, and that's because it was basically trained on materials.
• 1:06:27
  SPEAKER_02: And people said, like, well, it's gonna come from a company like Meta or Microsoft, so they will have done everything right.
• 1:06:33
  SPEAKER_02: So it can't be that they're that it's flawed, so that confirmation bias is then completely kicking in.
• 1:06:39
  SPEAKER_02: So that's something always question everything, I think is super important.
• 1:06:43
  SPEAKER_01: Yeah, yeah, there is a there is a I don't know if you know this guy, Elisir Yutkovsky, it's a really cool mathematician and AI researcher from the States, and he has a very cool like quote, one of my favorite ones, which is what do I know and why do I think I know it?
• 1:07:00
  SPEAKER_01: What do I know and why do I think I know it?
• 1:07:03
  SPEAKER_01: And this is the question that has been following me for the last 15 years through all my assumptions and ideas and opinions, because I would be always asking myself, What do I know?
• 1:07:15
  SPEAKER_01: And why exactly do I think that this is true or this is not?
• 1:07:19
  SPEAKER_01: And I think we should we should all do that, especially in development, because we're working with critical structures here, critical decisions.
• 1:07:27
  SPEAKER_02: Yeah, it is interesting how all it goes back to philosophers as well.
• 1:07:31
  SPEAKER_02: I mean, like uh there's probably a Greek philosopher that said exactly the same word, like 600 before Christ or something, and uh uh Locke in England or like uh the journal of Schopenhauer.
• 1:07:41
  SPEAKER_02: There are so many interesting bits in there, and it always feels like the like, oh, why should I talk about this old stuff?
• 1:07:46
  SPEAKER_02: We we keep trying to reinvent humans and not understand that we actually solved that problem a long time ago.
• 1:07:53
  SPEAKER_01: Yeah, yeah.
• 1:07:54
  SPEAKER_01: Unfortunately, it would be really cool, you know, if, for example, when you are born, you are being uh like all of this knowledge, like humanity technology, is being installed in your brain, and you don't need to repeat it multiple times in a row.
• 1:08:08
  SPEAKER_01: But unfortunately, it's not the case, and we need to learn, we need to study, and we need to repeat those things.
• 1:08:13
  SPEAKER_01: I was actually also before uh presenting this talk, actually, at the we are developers, I was slightly nervous because I was thinking that maybe it's something incredibly obvious.
• 1:08:22
  SPEAKER_01: Maybe people already know that, and developers, yeah, they're so intelligent, they know so much.
• 1:08:27
  SPEAKER_01: And then after the talk, lots of people came to me and they were like, Yeah, I've never heard about it.
• 1:08:32
  SPEAKER_01: Like, how do and like so it's so interesting?
• 1:08:35
  SPEAKER_01: We have to repeat those things.
• 1:08:37
  SPEAKER_03: I do think people know that they people have this idea that they are different, though, you know, it's this idea that like everybody else has different things.
• 1:08:44
  SPEAKER_03: Of course, but I don't because I'm aware of them.
• 1:08:46
  SPEAKER_03: Yeah, and I yeah, what's interesting?
• 1:08:48
  SPEAKER_03: I I did it with a load of uh students that were like 18 years old when I used to be a teacher, and it was fascinating because they they were all, like you say, completely kind of unaware of that even these things could exist, but they were definitely not aware that they had them, but they were really good at recognizing them in other people, and I and and it's kind of a way of saying, like, it's funny, isn't it?
• 1:09:05
  SPEAKER_03: We all sort of have this tone of like holier than now, or like they've got flaws, but I don't.
• 1:09:10
  SPEAKER_03: But out loud, we'll all say, Yes, we know we need to kind of acknowledge our flaws, but it's like what you said, Chris, we we kind of don't like to because it's a little bit painful or embarrassing, or suddenly we look at ourselves differently.
• 1:09:22
  SPEAKER_03: Um, and I think actually the one thing that was just sort of popping into my mind the whole way through this was um whether you think about anything, any other kind of context, whether it be like playing in a band or just doing anything in a group, playing in like a football team or whatever, the the the combination of those people and the diversity of their skills and their their drawbacks as well, the things that they're not good at, and that somebody else can uh help teach them or like fill in those gaps.
• 1:09:45
  SPEAKER_03: That's like the beauty of collaboration.
• 1:09:47
  SPEAKER_03: I think that's like the best bit about building things or making something with someone else.
• 1:09:51
  SPEAKER_03: Um and now we're doing so much, as we said, with AI on our own, in these silos, uh accompanied by just the LLM, that uh it kind of gives us more credence to like why we need to break out and actually collaborate with people, human beings, um, but also being kind of honest with ourselves about our drawbacks, that we're not all we're not all perfect.
• 1:10:09
  SPEAKER_03: In fact, none of us are, you know.
• 1:10:11
  SPEAKER_01: Yeah, yeah, exactly.
• 1:10:12
  SPEAKER_01: It's it's it's what uh yeah, sorry, yeah, it's what game uh game theory basically says that if we collaborate, we win.
• 1:10:21
  SPEAKER_01: And that's why it's so um slightly like it's slightly depressing for me that now with AI, lots of people work uh by themselves.
• 1:10:30
  SPEAKER_01: You know, we don't have big teams anymore because yeah, I can do this feature in five minutes, I don't need anyone.
• 1:10:36
  SPEAKER_01: And after we we basically are being cooked in our own biases, in our own heads, in our own assumptions and solutions, and we don't have a chance to be a team and to play football, let's say, you know, sort of terms, yeah.
• 1:10:52
  SPEAKER_02: It's interesting when when teams become successful, like when the football world cup and the Spanish team won, and people said like it was just because they had no they had no superstars, they were a team, they actually they all covered each other, and and that was great to see that this is still a successful point there.
• 1:11:08
  SPEAKER_02: I also found that by by by learning about biases and actually challenging your own, you actually learn much better because you actually don't fall into the trap of having that one resource, but then we we now go into a world where we don't have search results in Google anymore, but we get an AI overview, which is probably a biased overview.
• 1:11:28
  unknown: Yeah.
• 1:11:28
  SPEAKER_02: What do you think about that?
• 1:11:30
  SPEAKER_02: Is that something are we are we teaching people to be biased by just giving them the one answer to rule them all?
• 1:11:36
  SPEAKER_01: No, of course we shouldn't be doing that, because it's dangerous, and we know we all saw this movie and we read this book.
• 1:11:44
  SPEAKER_01: Uh, but um it's it's exactly the same.
• 1:11:47
  SPEAKER_01: The same approach of uh the same approach of being a scientist.
• 1:11:50
  SPEAKER_01: If you have this one answer, you should ask the question again, but from the different perspective, because this answer could be incredibly polished and it could feel like it covers everything that you have thought about, but it's not, and it can lose something.
• 1:12:07
  SPEAKER_01: And what we need to do is just think to ourselves what could break this assumption, what could break this belief that I have, you know, exactly the same as scientists do.
• 1:12:17
  SPEAKER_01: And what is what's very unpleasant about that is that we is the more we do that, the more often we see ourselves wrong.
• 1:12:26
  SPEAKER_01: The more often we see that, yeah, I was actually wrong.
• 1:12:29
  SPEAKER_01: I did this stupid thing and this stupid thing, and I shouldn't have done that.
• 1:12:34
  SPEAKER_01: But the more we realize that, right?
• 1:12:36
  SPEAKER_01: The more we see our mistakes and our flaws, the better we become as humans, and the better we become as team players and in general as developers.
• 1:12:46
  SPEAKER_02: Cool.
• 1:12:47
  SPEAKER_02: Well, we have a we have a thing called uh uh also we started doing a thing where I tell people uh uh I have a new feature on the show called Dan versus German.
• 1:12:57
  SPEAKER_02: Dan is always fascinated by how many German words we have that are sort of that are really long and mean a lot of things.
• 1:13:02
  SPEAKER_02: And I've got three for him today.
• 1:13:04
  SPEAKER_02: One of them just came to me.
• 1:13:06
  SPEAKER_02: Uh, there's a German saying that uh if when it comes to biases and it comes to being limited, you have to uh to wrestle your internal pick dog.
• 1:13:14
  SPEAKER_02: And it's called the inneren Schweinehund bekämpfen.
• 1:13:17
  SPEAKER_02: And this is like a fictional animal that basically does everything wrong, and that basically means you just keeps you from doing things.
• 1:13:25
  SPEAKER_02: So that's basically the one that stops you from being better, and that's your internal pig dog, the inner Schweinehund.
• 1:13:32
  SPEAKER_03: That's so good.
• 1:13:32
  SPEAKER_03: I don't this I don't know which part of that I love the most, but I have to say just pig dog.
• 1:13:36
  SPEAKER_03: I mean, I don't know, it's just brilliantly German.
• 1:13:38
  SPEAKER_03: I don't know what it is, it's great.
• 1:13:39
  SPEAKER_03: It's great.
• 1:13:40
  SPEAKER_05: Yeah, it's fantastic.
• 1:13:42
  SPEAKER_02: Another great one for you is uh is Stuben Kieger, which means uh living room tiger, and that's uh a name for a cat.
• 1:13:53
  SPEAKER_03: Okay, okay.
• 1:13:56
  SPEAKER_03: I love that.
• 1:13:57
  SPEAKER_03: I don't know what it is again.
• 1:13:58
  SPEAKER_03: It's is it is it is it meant to be a kind of like takes you a second to figure out what it is, or is it you know, like a cockney rhyming slang thing where it's like apples and pears mean stairs, you know?
• 1:14:09
  SPEAKER_03: It's kind of that's just brilliant.
• 1:14:11
  SPEAKER_03: Wow, living room tiger, and the last one is curious.
• 1:14:15
  SPEAKER_02: The word being curious in German means neugierig, which means greedy for new.
• 1:14:21
  SPEAKER_02: Ha!
• 1:14:23
  SPEAKER_03: Lovely, lovely.
• 1:14:24
  SPEAKER_02: That's also like explaining exactly what curious is, you know.
• 1:14:27
  SPEAKER_02: It's just and it's one word again.
• 1:14:29
  SPEAKER_02: It's brilliant, yeah.
• 1:14:31
  SPEAKER_01: It's the same in Dutch, actually.
• 1:14:32
  SPEAKER_01: Yeah, it's the same in Dutch.
• 1:14:33
  SPEAKER_01: News girl.
• 1:14:35
  SPEAKER_01: Also, like wanting more knowledge, yeah.
• 1:14:37
  SPEAKER_01: But how would you say a bias?
• 1:14:40
  SPEAKER_01: How would you say a bias in German?
• 1:14:43
  SPEAKER_02: Bias.
• 1:14:45
  SPEAKER_01: Bias, oh really?
• 1:14:46
  SPEAKER_01: Because in in in Dutch.
• 1:14:50
  SPEAKER_02: I can't remember.
• 1:14:51
  SPEAKER_02: There is a specialist term.
• 1:14:52
  SPEAKER_02: What's it in Dutch?
• 1:14:53
  SPEAKER_01: Yeah, in Dutch it's for ingenomenheit.
• 1:15:00
  SPEAKER_03: Exactly.
• 1:15:01
  SPEAKER_03: What does that what does that break down into?
• 1:15:03
  SPEAKER_03: That or is it just translate to bias?
• 1:15:04
  SPEAKER_02: It sounds like a for Eingenommenheit means like you already have uh you already have a conclusion for something before you actually go there.
• 1:15:13
  SPEAKER_02: You're uh you're you're predetermined to have a wrong outcome.
• 1:15:17
  SPEAKER_01: Kind of prejudiced, yeah, for the negative, yeah.
• 1:15:20
  SPEAKER_03: I think I think I know what it is.
• 1:15:22
  SPEAKER_03: German German is the equivalent of uh a coding language with the comments, it just tells you what it is as you go.
• 1:15:28
  SPEAKER_03: That's brilliant, because you'd have to go, what's bias?
• 1:15:30
  SPEAKER_03: I don't know what that is, where someone's just left you a comment saying you've already got the conclusion, that's amazing.
• 1:15:34
  SPEAKER_05: Fantastic.
• 1:15:34
  SPEAKER_02: Or when you could like Java when you had like Java and C developers that had like method names that this long, and you basically, as a Java guy, you say, like, do that.
• 1:15:44
  SPEAKER_02: It was your function name, like you know, do things.
• 1:15:50
  SPEAKER_02: Cool.
• 1:15:51
  SPEAKER_02: Well, thanks very much, Kate.
• 1:15:52
  SPEAKER_02: Uh, we're gonna link your talk after this as well to basically for people to get more.
• 1:15:56
  SPEAKER_02: Like, uh, do you have any resource where we can uh where there's a great introduction to biases that people can take a look at?
• 1:16:05
  SPEAKER_01: Uh so I would say that obviously there is uh Wikipedia and there is lots of information about that, and there is a great picture which is called uh I actually also have it in my talk, which is called Cognitive Biases Codex.
• 1:16:19
  SPEAKER_01: But what I would recommend people to start with if they want to have some fun uh engagement with uh biases, I would recommend the book by Elisa Riutkovsky, the same guy, um, which is called Harry Potter and the Methods of Rationality.
• 1:16:35
  SPEAKER_01: Uh, this is basically the book in the same universe as original Harry Potter books are, uh, but with the condition that Harry was born in a loving family of scientists, and he's incredibly um knowledgeable there.
• 1:16:51
  SPEAKER_01: He knows a lot about biases and how all of those experiments were like proceeded with.
• 1:16:58
  SPEAKER_01: And he shared lots of stuff, but with magic, who in the same universe with lots of additional interesting details.
• 1:17:05
  SPEAKER_01: I like I highly recommend.
• 1:17:07
  SPEAKER_01: I read this book three times, and I didn't read original Harry Potter ones, uh, because this one for me feels like you have to read it.
• 1:17:16
  SPEAKER_01: It's fantastic.
• 1:17:20
  SPEAKER_03: It's fun fiction, though I think, as well, right?
• 1:17:22
  SPEAKER_01: Yeah, it's fine fiction, yeah.
• 1:17:23
  SPEAKER_01: Exactly, exactly.
• 1:17:24
  SPEAKER_01: And it's it's it's fantastic.
• 1:17:26
  SPEAKER_01: Yeah, and of course, if you want to something deeper, you can also read the book, which is also by the same guy, which is called Rationality from Artificial Intelligence to Zombie, where he explains in more details about statistics, about how we make decisions, how we come to a conclusion as humans, including all of the possible biases and additional things that in our brain, but not from biological perspective and neurobiological, but from just the thinking processes, you know, like how to be a rational human being and what it means.
• 1:17:56
  SPEAKER_02: What about the uh for uh uh uh thinking fast and slow one?
• 1:18:03
  SPEAKER_01: Yeah, it's a good one, but I stopped recommending it because it feels like soap culture for to me these days, you know, like it feels like everyone knows this book.
• 1:18:12
  SPEAKER_01: So I I I try to come up with something more uh unique, I would say, in this sense.
• 1:18:18
  SPEAKER_01: But yeah, it's also it's it's a really good book, yeah.
• 1:18:21
  SPEAKER_02: Cool.
• 1:18:22
  SPEAKER_02: Very clever to to to to piggyback on Harry Potter as well.
• 1:18:25
  SPEAKER_02: That success, isn't it?
• 1:18:28
  SPEAKER_01: Yeah, yeah, highly recommend, guys.
• 1:18:30
  SPEAKER_01: Like, highly recommend.
• 1:18:31
  SPEAKER_02: Well, I was thinking about that.
• 1:18:33
  SPEAKER_02: The whole thing in Harry Potter where you have to have a patrolist charm to actually defend yourself against a Dementor.
• 1:18:38
  SPEAKER_02: If a Dementor is an evil being that is actually sucking souls, just hide behind any redhead person.
• 1:18:47
  SPEAKER_03: And you can make that joke, you know, if we said it, I'd be canceled, but yeah.
• 1:18:54
  SPEAKER_01: Oh my god, yeah, that's true, that's true.
• 1:18:57
  SPEAKER_01: Um, yeah, Kerstman from South Park would be agreeing with you right now, would be nodding.
• 1:19:02
  SPEAKER_01: Yeah.
• 1:19:05
  SPEAKER_02: Cool.
• 1:19:06
  SPEAKER_02: And with that, that's the end of today's developers live.
• 1:19:10
  SPEAKER_02: Thanks very much, Kate.
• 1:19:11
  SPEAKER_02: This was wonderful.
• 1:19:12
  SPEAKER_02: Uh, we're gonna talk, we're gonna link your talk so you can people can learn more about it.
• 1:19:16
  SPEAKER_02: Where can they find you if they want to know more about you?
• 1:19:19
  SPEAKER_01: Uh yeah, LinkedIn and Instagram.
• 1:19:24
  unknown: Cool.
• 1:19:25
  SPEAKER_02: Okay, well, Dan, thanks very much, Kate, very much.
• 1:19:28
  SPEAKER_02: Uh Kate very much.
• 1:19:30
  SPEAKER_02: Kate, thanks very much.
• 1:19:31
  SPEAKER_02: Uh this was We are developers live for today, and now we're off to do spooky things.
• 1:19:36
  SPEAKER_02: See you guys soon.
• 1:19:38
  unknown: Thank you.
• 1:19:39
  SPEAKER_05: Bye bye.