328: I (Git) Blame You

I (Git) Blame You

[00:00:00] Intro: Tired. So tired, over tired.

[00:00:04] Christina: You are listening to Overtired. That’s right. We are back. I’m Christina Warren, joined as always by my friends, Brett Terpstra and Jeff Severns Guntzel. Boys. How are you? It’s been a while.

[00:00:15] Brett: It’s good to be back.

[00:00:16] Jeffrey: Yeah, good. You’re going to hear birds. Maybe because I refuse to shut my windows for audio quality because it’s Minnesota and it’s warm, and so let’s do

[00:00:26] Christina: Absolutely.

[00:00:27] Brett: directional mic. It’s working

[00:00:29] Jeffrey: Yeah, that’s true. That’s true. I also have a cat that has started walking across my keyboard. I, I call him the intern now because he sent an email and deleted another, so

[00:00:37] Mental Health Corner: Pets Edition

[00:00:37] Brett: You guys want a, a Yeti update?

[00:00:39] Christina: How How’s Yeti doing?

[00:00:40] Jeffrey: yeah, the old cat. How’s the old cat?

[00:00:42] Brett: So like yesterday I took him, he’s getting these shots. I can’t remember what they’re called, but they are supposed to help with like mobility and aging cats and, and they work for ’em. Um, But yesterday when I was in, I mentioned some specific problems he was [00:01:00] having and they were like, oh, this combination of him aging, dropping weight, and then these like basically stool issues, um, could be a really bad sign.

[00:01:12] Brett: So we would like you to make an appointment to uh, do some, do, do a consultation, do a full physical, and I paid like $400 for labs. And then we had to go in today and I was literally expecting the worst because like two cats in a row, we took Clovis in because his breast smelled bad. And they were like, oh yeah, it’s oral cancer.

[00:01:35] Brett: He has two weeks to live.

[00:01:37] Christina: my

[00:01:37] Brett: And you know, like this was a shock. And then we took Finnegan in because his meow had changed and we were concerned about like maybe something in his lungs or something. And he’s nine months old and they’re like, He, he’s got two weeks to live. So I have this like, fear of these appointments, but we went in today and they’re like, oh my God.

[00:01:58] Brett: Yeah. E even [00:02:00] Yeti’s, uh, kidney illness. Ha the scores have come down. Everything’s looking great. You guys are doing a great job. Uh, we’re gonna, we’re gonna treat a bladder infection and we’re gonna put him on some more meds for his, uh, his runny stool. But yeah, they’re like, you’re doing a great job. This, this cat’s doing great for a 19 year old cat.

[00:02:21] Brett: And I was like, oh. So relieved. It

[00:02:23] Jeffrey: amazing.

[00:02:24] Brett: it was because I was crying last night. I was preparing myself for the end of life, right? And so, like, I kept like, breaking down and I was like crying in front of Elle, just like trying to like deal with my, like, it’s time. We all know it’s time, but it’s mortality and it gets me.

[00:02:45] Brett: And then today was such a relief. I have a little more time with my boy.

[00:02:49] Christina: Yeah, that’s really good. That’s really good. I know. Yeah. And I know what you mean. Like having like that fear of like going to the doctor and hearing stuff because you’ve, that that’s the only experience you’ve had and like it sets you up. It’s, [00:03:00] you

[00:03:00] Brett: Every, every pet I’ve ever had has died of cancer. Like there’s always this late stage cancer discovery and like with, with Emma, like we, we found out she had cancer and had to put her down the same day, like it all

[00:03:16] Christina: uh, that’s the worst

[00:03:18] Jeffrey: no wonder you’re pre crying going to the vet, you know?

[00:03:22] Brett: I’ve had some trauma. Yeah.

[00:03:24] Jeffrey: Yeah. Wow. I’m so happy.

[00:03:27] Brett: you. Thank you.

[00:03:28] Christina: That’s wonderful.

[00:03:29] Brett: That might, that might even count as my mental health corner update.

[00:03:32] Jeffrey: There you go. You’re, are you stepping out of the corner? I.

[00:03:36] Brett: That’s me in the

[00:03:37] Jeffrey: Oh no. Oh,

[00:03:41] Christina: Uh,

[00:03:41] Jeffrey: It’s gonna be stuck in my head.

[00:03:43] Brett: I know, right? Have you ever seen the cat version of losing my religion where it’s just photos of cats and it’s like they illustrate like every line, like there’s literally a cat sitting in a corner and then there’s a cat, cat under a light

[00:03:56] Christina: no I haven’t.

[00:03:57] Brett: that’s me in the spotlight. And then [00:04:00] a cat in front of like a crucifix and it’s,

[00:04:03] Christina: really funny.

[00:04:04] Brett: it’s amusing.

[00:04:05] Jeffrey: So there’s a, there’s a great story from a, a wonderful photographer based in Minneapolis. His name is Alex. So, and he’s this like kind of international photographer. He makes these, he’d uses like large format, um, cameras, and he makes these just stunning portraits. And at a certain point he started being asked to do things for like the New York Times Magazine or other things like that, like kind of big market stuff.

[00:04:29] Jeffrey: And, and he was, he was told that Michael Steppe wanted to, um, wanted him to take his photo and, and he was at a point where he was just like not feeling, uh, this whole idea of like applying his art to celebrities. And so the photo he suggested, and ultimately the job was killed because of this, was that they, they meet in New York.

[00:04:48] Jeffrey: And that Michael Stipe stand two blocks away and they take this gorgeous, cuz the, the large format camera is like, every detail of what’s in it is beautiful. So it would be a really compelling [00:05:00] photo. But Michael Stipe would be two blocks away, or one block probably. And, uh, so that, that got killed. But anyway, the, the r e m story always reminds you of that.

[00:05:09] Jeffrey: I think it’s just a fun story. It’s not even snotty, it’s just like, he was like legitimately that’s just where he was

[00:05:14] Brett: I love that idea. I feel like, yeah, if I were gonna have, I would love that picture of me to be like part of something larger and not have it be all about me. I, I would, I would and, and I would think Michael’s sip, like from what I know of him.

[00:05:29] Jeffrey: You would think you’d be into it. Yeah. But like I, the, the cool thing, I’ll link this in the show notes, is that he has a photo, I think he was using as reference, which is this photo of a monk in the woods and the monk is just way off. Um, it’s just, it’s just the most amazing portrait. Anyway, I’ll pick up on the theme, which is just that I, I, I have this cat sitting next to me.

[00:05:50] Jeffrey: His name is Murphy. He’s totally my best friend and my favorite person. And when he was a kid and he ate my Zoloft, that almost died. Um, and I often think [00:06:00] back to that and I get this like, enormous wave of grief. Uh, even though he’s fine, he is three years past it, it’s fine. But, uh, I, it just. I mean, so here’s like a, another cat related mental health check-in.

[00:06:14] Jeffrey: I’ve started, I’ve started having a bed on my desk, like right to my right, which looks out a window. And my two cats, which are, they were they’re siblings. They were found, um, in a barn alone when they were little kittens. And so they, they’re like constantly together, constantly snuggling. So they actually, at this bed is really only big enough for Murphy, who’s the big boy.

[00:06:34] Jeffrey: And then my other cat looks like the runt of a raccoon litter. And, and they sit in here and they make a little, like little fur pile and they snuggle while I’m working, like almost all

[00:06:43] Jeffrey: day. And so if I’m having meetings that are stressful, I have one hand on this like giant fur pile, you know, and you can kind of feel the purring and feel the, the breathing.

[00:06:53] Jeffrey: And

[00:06:54] Brett: Our cat, nobody, uh, grew for a year and then just stopped growing. [00:07:00] So now she’s over two years old and still it’s the size of like, maybe even less than a one year old cat. She just, she’s a tiny, she’s a tiny cat. She’s a runt. She’s a runt of a, what’d you say? A raccoon litter

[00:07:13] Jeffrey: Yeah, that’s what, that’s what my other cat looks like and acts like. Does that name nobody, did it come from the movie Dead Man, or did it come from

[00:07:21] Brett: came from the graveyard book by Neil Gaiman.

[00:07:23] Jeffrey: awesome?

[00:07:24] Christina: so my, I guess, uh, in my updates, so this is a couple weeks old now, but my, uh, my sister’s dog, boo Bear, who we love, he’s definitely my sister’s first son. He, uh, he’s a poodle. Um, he’s whatever the, not the tiniest one is, which I guess is toy. He’s mini poodle, uh, or miniature poodle, whatever. But, but he, um, and, uh, and his twin brother, uh, P bear are, are not, um, and, and Pbe belongs to my mother-in-law.

[00:07:49] Christina: But, um, boo Bear has cataracts or had cataracts, um, And, and it got to the point that like, it was really, really bad and he was basically completely blind and he had surgery last week [00:08:00] and he went through it pretty well. They wanted to add the lenses, but the vet basically said that she didn’t wanna keep him under that long cuz he doesn’t do super well with being under anesthesia.

[00:08:10] Christina: So he didn’t have lenses put in, but the cataracts are at least removed. So hopefully he will at least get some of his, you know, vision back. Like, I think he’s gonna be able to see things from across the way, but he’ll still have a hard time with things up close. But I’m hopeful that that will at least improve some of his quality of life.

[00:08:26] Christina: Cuz it was really, really sad to see him not like he, he knows he’s been staying with my parents. Um, because my sister’s gone a lot during the day and he knows the layout to most of their house, but like, not all of it, like, he got stuck in like the bathroom in the bedroom that I stay in. It’s, it’s hard seeing, um, seeing animals deteriorate.

[00:08:43] Jeffrey: Sure is.

[00:08:45] Brett: All right. Well, we have some topics.

[00:08:47] Apple Savings

[00:08:47] Jeffrey: Well, do you wanna talk about the um, apple savings account? Cause I know you really wanted to hit on that.

[00:08:53] Brett: I just feel like everyone needs to know about this. So if you have an Apple card, you can [00:09:00] now, uh, sign up for an Apple savings account also through Goldman Sachs. And it comes with, and I assume this is true for everybody, it comes with like a 4.15 interest rate, and there’s no minimum balance.

[00:09:15] Brett: There’s no minimum deposit, there’s no yearly fee. It is a far better savings account than I can get through my local credit union. I moved most of my savings at this point into this Apple savings account, and instead of, instead of your cash back from your Apple card going to an Apple Cash account, it can now go into an Apple savings account, which is earning four plus percent interest.

[00:09:42] Christina: I was gonna ask you about this, cuz I’ve been looking at that cuz when they sign, when they uh, you know, introduced it, I, I signed up almost immediately cuz I had um, I think about a thousand dollars in Apple cash for various things. My typical thing is that I get like my, my Apple cash back. And then I usually let it stay in there, just kind of in [00:10:00] perpetuity.

[00:10:00] Christina: And then like, you know, once it reaches a high enough balance, I pay for something with it or I, I transferred to my, I transferred to my bank account, is usually what I do. And in this case, I just hadn’t, and, and I was wrong. It wasn’t, um, uh, it wasn’t a thousand dollars, I think it was 500 cuz I transferred $2,000 to my bank account.

[00:10:16] Christina: So I, I’d had like, I had like 2,500 in there and then I was like, I was like, I was like, I’m gonna put. Two, two. That was like back in my, in my, um, my checking account, which was probably dumb because what I was gonna ask you is cuz I have been considering like, I think that they let you put in like 20 grand, like on a, on a, a seven day period.

[00:10:34] Christina: But like, how hard is it, I guess to get money in and out? Because I wouldn’t be opposed to putting, like with that 4% thing, like you said, that’s better than a credit union. It’s better than I can get with Bank of America and they give me some pretty decent rates because, um, of, of how long I’ve been there.

[00:10:49] Christina: But like how, uh, how easy is it to get stuff in and out? Because if that’s the case, like I could see myself putting, you know, like, like 20, $40,000

[00:10:57] Brett: like your, whatever you pay your [00:11:00] Apple card with, uh, it’s automatically connected to the same bank account. Um, putting money in is seconds getting money out. Like, uh, I transferred money out of Apple savings for Yeti Vet Bills, and it took about 24 hours for the money to show up in my bank’s checking account.

[00:11:20] Brett: So I wouldn’t say it’s difficult, it’s not instant, but as far as bank transfers go, it’s not two to three days, it’s a day.

[00:11:28] Christina: Okay, cool.

[00:11:30] Crazy-Ass Google

[00:11:30] Brett: As long as we’re covering my quick hit topics. also wanna mention that Google’s tap level, domain zip is a horrible idea. Like right now, if you go to malicious content.zip, they have registered, I think there’s a blog called Malicious Content about, uh, Trojans and viruses and, and, and cybersecurity.

[00:11:54] Brett: Um, if you go to malicious-content.zip, Firefox will [00:12:00] download a payload.

[00:12:01] Christina: will Chrome and then, and what And, and what’s inside that is A P D F, which is also problematic, right? Because then, which honestly I have to say this is brilliant on the cybersecurity people who put that together because not only is the payload a zip file, which is bad enough, but then what’s inside the zip file as a P D F file.

[00:12:17] Christina: So these are two things that have been very, very exploited by, by viruses and macros and all kinds of other things. Like we’re, we’re Mac users, we’re typically more immune to these things. Um, not universally. Definitely people can create macro trojans, but usually not in, in PDFs and stuff, but like, this is, this is how, this is how you get ants as they say.

[00:12:37] Christina: Like, if you, do you want ants? This is how you could, ants

[00:12:40] Jeffrey: Yeah. A PDF inside a zip file is kind of like the turducken of, um, you know,

[00:12:45] Christina: Yeah. Like, like, like, like, like it was a, it was a doc file, not even Doc X, like doc, like, that would be like, like adding like, uh, uh, you know, the next level of, uh, of, of, of turducken ness. Like that would be like adding a, another, another piece of lobster or something.[00:13:00]

[00:13:00] Brett: what do you think Google was thinking when they made a.zip domain?

[00:13:04] Jeffrey: confusing cuz obviously it’s the first thing you think is like, this is no

[00:13:07] Brett: Yeah, yeah.

[00:13:08] Jeffrey: super smart. So why did they knock that concern

[00:13:12] Jeffrey: down?

[00:13:12] Brett: feel like, I feel like somebody would’ve said as a top level domain, this is, this is ill faded. Like they have AI experts, they have ethics experts.

[00:13:23] Christina: have massive security teams. What are you talking about? Like they, they, they literally have like one of the best bounty teams where they find other people’s vulnerabilities and report them to them. Like they literally, like Google has some of the most talented, you know, engineers working for them.

[00:13:38] Christina: We know that. But they also have very, very smart security people cuz they have to, and it’s so dumb. And the thing is, is like they also bought, uh, dot MOV domains, which is also a problem because QuickTime files.

[00:13:49] Brett: like a exe domain.

[00:13:51] Christina: It really is. I mean, and, and, and look, I know that there are purists out there. Cause I’ve seen these comments on various things like, well Calm was, was a file extension back in the eighties [00:14:00] and this and that, and I’m gonna be like, okay, fuck off.

[00:14:02] Christina: Neck Beard. No one had the web in the eighties, first of all did not exist. The people who had access to those things were like people who worked at universities and institutions or very, very eager hacker kids. And even then they couldn’t afford to buy domains. So it was, to me it’s just not even a comparable situation.

[00:14:22] Christina: Like the.zip thing is really bad because there are a lot of people who will have things hard coded in. And, um, I saw some, um, post who basically showed, like, because of some of the, the various, uh, unicode, uh, fuckery things you can still do with domains that, you know, the, uh, ICAN has still refused to address for years, despite being called on it for like literally years and years and years.

[00:14:44] Christina: You can make two files look identical, where one would take you to like a GitHub repo. They would have a zip thing and one would look the same. And if you clicked on like, okay, which one of these is legit? And like, if you clicked on one of them, it would like take you to like, there’s nothing here. And if you clicked on another one, it would download a [00:15:00] zip file.

[00:15:00] Christina: But one of them is not coming from a GitHub domain. It’s coming like it’s, all of this has been spoofed and, and it’s just like, like, you know, Unicode fuckery, this is really, really bad. But, but because of the, the way the, you know, they, they can, you know, modify this stuff with the.zip stuff. It’s just, to your point, they should know better.

[00:15:18] Christina: And, and this isn’t the first time they’ve done stupid shit with their, um, top level domains. Like, do you remember when they bought, they bought.app and then they bought.dev and.dev was, was, um, problematic for a lot of people because, um,

[00:15:33] Brett: testing domains. Yeah, I, I

[00:15:35] Christina: exactly. Exactly. So,

[00:15:36] Brett: of my, all of my local host domains. I had to change, instead of bt.dev to test my local website, I had to change to dev.bt and use an

[00:15:48] Christina: Right.

[00:15:49] Brett: t l d for all of my extensions. Yeah.

[00:15:52] Christina: Exactly. And, and so, so they’ve, they’ve done this before and like there’s a part of me that I’m like, okay, maybe if they had bought it preemptively, because you [00:16:00] know, they, they take on all these top level domains. It’s like, okay, well maybe Google did this is for a good reason and they’re not gonna roll it out to people and, and protect the internet.

[00:16:07] Christina: But no, they’re just like, no, well, we’ll we’ll sell ’em to you. Why not? It’s like,

[00:16:11] Brett: Oh,

[00:16:12] Christina: it’s like, shut up.

[00:16:13] Jeffrey: It’s amazing.

[00:16:14] Brett: do know. Don’t, don’t be evil. Yeah.

[00:16:18] Christina: Yeah. That, that was a long time ago.

[00:16:19] Brett: was a long time ago. It seemed so quaint in retrospect.

[00:16:22] Christina: It really does. And you know what’s funny is like there are a lot of people I’m most, I’m absolutely including myself in this, who really gave Eric Schmidt a lot of shit when he was c e o and chairman. And I don’t, not saying that that was misguided cuz definitely like the guy is, You know, weird and whatever.

[00:16:39] Christina: But if I compare him to like, what happened when he left, and then especially like Sundar, who just seems like the, the perfect combo of like incompetent and like aloof. Eric Schmidt at least had some balls. Like Eric Schmidt to his credit, pulled out of China, full stop. He said, we are making a business decision.

[00:16:58] Christina: We will not operate [00:17:00] in China. Period. Full stop. And then it was years after he left when they were like, oh, well this is a really big market. We really, we have to find a way to sort of, you know, operate, but not really operate. They kind of reneged on that. But like, I still, in my opinion, that was one of the most standout standup like business moves I can ever recall any company taking because no other tech company made that decision.

[00:17:21] Christina: And that was Eric Schmidt. So like, I’m, I’m sorry, Eric Schmidt. He, cuz he was the one who created Don’t Be Evil. I was like, I, I’m sorry for dogging on you. You, you suck, but you are better than, than our current crop of tech CEOs in re.

[00:17:35] Brett: In retrospect.

[00:17:36] Jeffrey: You got me thinking on the don’t be evil front. Like it’s not too easy to create a shared definition of what evil meant at the time. Um, but certainly you can, you know, sketch one together and I, I wonder how different our definition of evil used in the sentence Don’t be evil is, has changed between that time and now.

[00:17:57] Christina: Oh, that’s a great point. You know what? I bet, I bet in [00:18:00] some ways it’s gotten like things that we would not have considered evil. We do. And I bet in, in some ways that it, it, it’s reversed, right? Because Google, like a lot of those tech companies, and I’m sorry we’re going on a tangent, but like, I’m not talking like the libertarian like party like of Silicon Valley companies that were much more like small l libertarian.

[00:18:17] Christina: Like I’m not talking about like, you know, the, the, the people who, who claim those ideals. I’m not talking about like Connor from succession. I’m just talking about people who created the internet archive and the, you know, um, electronic frontal foundation and other kind of, I guess in some ways would kind of like be like liberal, like anarchists, right?

[00:18:35] Christina: Like they had like very specific ideas around freedom of speech and around accessibility of things and, and around like stopping, you know, government, um, inter intervention and things. And I think in generally in trying to kind of do the right thing as they define it, right? We feel it this way and we can do it.

[00:18:52] Christina: And that’s just not really the ethos. Anymore. Like that was not, I’m not saying that it was, it was absent profit motive, cuz it [00:19:00] absolutely wasn’t. But it wasn’t as tied into kind of the, the, the corporate greed cycle that we’ve had now. And, and so it’s, it, it, I I think in some ways things that they would’ve called evil, um, we now would maybe not feel as strongly about.

[00:19:15] Christina: Um, but in some things that they, they would be like, oh, this is fine. We’d be like, oh no, that is straight up evil.

[00:19:19] Jeffrey: Google itself.

[00:19:20] Christina: That, that’s an interesting topic. I wish somebody would write that. Yeah, well, exactly,

[00:19:26] Jeffrey: Yeah. Well also it was, I remember when I first read that, I mean in, in the actual days, uh, days of your, um, I remember thinking, oh, that’s cute. This, it seemed like kind of a cute statement. You know what I mean? Um, it didn’t seem like something that went through like, uh, several large committees cause they didn’t exist at that time.

[00:19:44] Christina: exactly. I was gonna say, they, they had so few people. It was, it was genuinely a startup. Right. And um,

[00:19:50] Jeffrey: I feel like there should be a Minister of Prophecy at all of these, um, tech companies so that when you’re creating these, these, uh, slogans or promises, someone can be like, um, just a second. We’re [00:20:00] gonna have to speak to the Minister of Prophecy.

[00:20:02] Brett: Notably, notably, no one since them has had a similar tagline.

[00:20:08] Christina: Yeah. I mean, I mean, the closest was, and it was obviously, um, different. It, it didn’t have like the moral, well, eh, a little bit, but like Facebook’s was like, move fast and break things right. And, and, and then they had to kind of drop that. But like that they are still dogged by that to this day. I think the two of them, I think that that was the lesson.

[00:20:26] Christina: Like, I don’t even know if you need a Minister of prophecies now. I think you just look at history and you go, everyone is going to use this slogan that you think is great and applies to your, you know, small, you know, few thousand, few hundred person company. It is going to define who you are when you become a trillion dollar company.

[00:20:46] Christina: Uh, and, and people are still going to like, hold it against you or, or a 500, you know, billion dollar company. If your Facebook, like, we’re, they’re gonna, you know, hold it against you in perpetuity forever. This will define you forever.

[00:20:58] Brett: I feel like the heat [00:21:00] death of the metaverse is a topic for a whole other show.

[00:21:04] Christina: Oh my God. Yeah. Yeah. That is whole other show. Imagine if you named your company after the metaverse.

[00:21:10] Jeffrey: They meant metadata.

[00:21:12] Christina: Hey, good segue.

[00:21:13] Jeffrey: Oh yeah,

[00:21:14] Jeffrey: that’s true.

[00:21:14] Brett: let’s, let’s opt not to talk about Facebook or Elon Musk. Let’s talk about Jeff’s foray into ai.

[00:21:23] mundaneGPT

[00:21:23] Jeffrey: Now there are two levels to that. One is sort of working chat g p t into my normal, mundane, everyday life, and the other is far more exciting work with the a p I and deeper stranger

[00:21:34] Jeffrey: work.

[00:21:35] Brett: see what we have time for.

[00:21:36] Jeffrey: My idea is that we talk about the sort of, uh, working something like chat g p t into your everyday life and we save this big philosophical AI self-portrait conversation for the next

[00:21:48] Brett: Yeah, that works.

[00:21:49] Christina: Yeah.

[00:21:49] Jeffrey: So I, I distinguish now because I do use chat g p t kind of all the time. I, I pay now so that I can use, so I can add, ask 25 things [00:22:00] of G P T four in every three hours. Um, goes fast. Uh, but anyway, um, so there’s that. And then there’s also sort of forays, thank you, Brett, into using the api, which is more of the topic for next week, but is has become much more interesting and exciting to me.

[00:22:19] Jeffrey: And it’s so easy to do. They have such great cookbooks, just to say. OpenAI has great cookbooks for doing all the various things, um, with their api. So anyway, I, um, I think we talked about this before and what, what I was doing at the time was I was experimenting with writing, um, scripts using Chap G p t,

[00:22:39] Brett: Scripts meaning automation.

[00:22:41] Jeffrey: writing a Python script to parse text and make basic changes.

[00:22:45] Jeffrey: And.

[00:22:45] Brett: so not like a movie script, which would make you a, a picket line breaker.

[00:22:51] Jeffrey: Early on. Yeah, that’s right. Early on, a friend of mine sent me something cause I had said something absurd to him in a text and he, using chat g p t, he turned it into a [00:23:00] Seinfeld kind of vignette where everyone gets a line. Um, anyway, so yeah, no, like writing a Python script to parse text or just whatever I was writing.

[00:23:08] Jeffrey: Uh, working with APIs, I would just say, you know, here’s the, you know, I use whatever service I’m using. Like, how would I do this in the da da da api? And it would tell me, and as long as the API hadn’t changed since whatever the date is, that is the cutoff for chat G P T or G p t, um, it would do a good job.

[00:23:25] Jeffrey: And I, and it, it kind of helped me with computational thinking because I would, you know, ask it a question. It would write an initial script. We would do a lot of stuff to fix it. But then I’d start over and I’d ask the question informed by. What I had learned in the previous chat. So I was like, the learning ai, right?

[00:23:44] Jeffrey: Um, and it would write, you know, it would write, uh, scripts. You put ’em in, you get errors, you put it, you know, give them the error and it’s deciphering errors, which is amazing. So anyway, so that’s that part. But I’ve also just been testing out kind of the various other [00:24:00] exercises of my, my, uh, profession. So I write a lot of records requests, and they’re usually state level records requests, which means that, you know, whatever state you’re requesting in, let’s say you’re requesting emails from the state probation, uh, department or something, right?

[00:24:16] Jeffrey: You have to know and cite. Nebraska’s, uh, public records law, which will tell you in which you then assert, you know, here’s how long people have to respond to you. You know, they have to respond to you by this amount of time. They can’t charge you in on this unless this right? And so I’ve got it writing public records requests for me, which was previously something I did using text expander.

[00:24:39] Jeffrey: So I just have all that kind of text framing. But what I’ve found is that when I ask chat g p t to write a record request based on where I’m going to send it, it writes a more interesting and more legally thorough. Records request, then I would’ve otherwise written. So that stuff’s been [00:25:00] interesting. And then, um, I’ve just been using it.

[00:25:02] Jeffrey: So, uh, on our kind of board of directors at the organization I’m part of. Every time we have a meeting and we have meeting notes, I run it into chat, G p t and ask for a summary as for a couple different kinds of summaries, a bulleted summary, a summary of the most important actions or topics, just to see how it works.

[00:25:20] Jeffrey: And it generally works quite well. And then I’ve also been feeding it my raw notes from phone calls or other kinds of meetings or interviews. And my raw notes, meaning like my trashy raw notes, like incomplete sentences, spelling errors, no real clear sense of when I’ve gone from one subject to another.

[00:25:39] Jeffrey: And it does a shockingly. Good job of inferring what is absent. Um, and, and kind of summarizing based on that. And then the, the other way I’m, I’m using it right now, and again, none of these, I’m never leaning on it as the main thing. Right? Everything has. My kind of review, careful [00:26:00] review, test it again, review whatever.

[00:26:03] Jeffrey: But, uh, here’s what is a concern that’s come up for me. Um, so if I am creating documents that will ultimately be historical documents for my organization, or if I’m doing research on, like, I have all this research on a ancestor of mine fought in World War I, and I have this like bulleted information, and I asked chat, g p t, like summarize this.

[00:26:25] Jeffrey: And then I said, now give me like the context, um, for this person’s unit. Now tell me what changed in the Midwest between 1917 and 1918, you know, when he left and came back. And, um, and I’m adding that into the document. And what I fear is that. The fact that this was generated by ai, which I’m noting and the fact that I am well aware that I can’t just depend on this might get lost down the road in, in the years that come, someone looks at this in 50 years and they’ll be like, oh, here’s interesting research on how the Midwest changed [00:27:00] in this period.

[00:27:00] Jeffrey: But in fact, it was generated by AI and I didn’t fact check it. I just wanted to do it to see what it would do. And like that actually brings up a big question I have for both of you. Like imagine that. Problem. Imagine it in many domains. Um, one example actually could be if I’m having summaries of meeting notes from chat G p t and it actually isn’t quite perfect, but I either don’t notice cuz I’m moving too fast or, um, or it just gets past me, right?

[00:27:29] Jeffrey: That becomes the record, right? Like, because the record, so if there’s one sentence off, it could one day be much more meaningful than that sentence is now. Right? It might mean I might blow by cause I’m like, well it doesn’t matter that it got that wrong. The rest of this stuff is great, right? But then I also think more in the case of being a researcher, being a reporter or whatever else, like if I imagine, you know, my story, uh, folders being in a historical society someday because I donate them just cause I feel like why not have all these Minnesota stories here?

[00:27:57] Jeffrey: How do we, how do we, what do we [00:28:00] do short of, I mean, beyond metadata, I guess to say, hey, point at it and be like, Hey, hey, hey.

[00:28:08] Christina: So, so that’s interesting. So it’s actually this great topic. I actually had a conversation, um, yesterday. With, um, a designer at Microsoft. Um, so Microsoft Build is next week, and I’m, I’m co-hosting and I was doing some pre-interviews, um, before the show. And this is actually gonna be a session, um, at, at Microsoft Build.

[00:28:25] Christina: Um, that was just recently out of the, to the schedule. And then Curtis and I are going to be talking, um, afterwards or, or before, I’m not really sure on the timing, but he, he’s a designer who’s been working on, um, basically thinking about, okay, what is the design language and what are the things that we need to build into these large language models and these interfaces, a, to help people use them?

[00:28:46] Christina: And then b, exactly what you were talking about to like, let people know about sometimes these things hallucinate sometimes the things that these things output are incorrect. Like what, what do we put into place there so that people don’t become overly reliant? [00:29:00] On these results. Exactly. To your point. And so I do feel like metadata is definitely part of it.

[00:29:05] Christina: Right. I think that’s a big and important part of it. And, and hopefully metadata can persist across file formats and generations and technologies and whatnot because, uh, that, that’s always a concern, you know, that that stuff can get stripped or, or lost or whatnot, and then you lose all context. Um, but I think there’s another part of it too, which is, and I hadn’t, and I have to admit, I hadn’t, I, I think I’d read the blog post that his team put out, um, uh, like a month or two ago about how they were approaching designing, um, for, for lawyers language models.

[00:29:34] Christina: But I, I, I hadn’t really, um, it hadn’t really fortified in my mind the same way, which was, okay, what are the design decisions that we need to do to let people know what’s up and what the truth is? And I feel like, you know, Even like in the notes, like in addition to having metadata, like it might need to be something where something is just called out, right?

[00:29:55] Christina: Like, might have to be that explicit being like, this is from, this was [00:30:00] generated from AI and has not been fact-checked. Like, but we ha But I think we have to think about everybody who’s designing these systems, how are you informing people and, and ensuring that the correct context is there because it’s so close, so often enough that it is incredibly easy to just become reliant on the results.

[00:30:19] Christina: And that’s the same thing that happened with Wikipedia in the early days, right? Like early Wikipedia. Early Wikipedia was garbage. It’s a lot better now than it was, but it was garbage. But we had access now to more information than we’d ever had access to at any previous time. So it was really easy for people to overly rely on it.

[00:30:36] Christina: And then I think people went and over-indexed the other way, which was like, you can never cite Wikipedia as a source. And I’m like, okay, maybe not. But you could take some of the sources that they cite, right? Um, and, and I think that, I think that, so I think that might be part of it too, which is I think that we have to do a good job of tagging things.

[00:30:53] Christina: I think we have to do a good job of in the products themselves, making clear what’s happening. And then I think there’s another [00:31:00] UI aspect where I think maybe some of these places, when they’re coming up with these claims, they have to add in footnotes as part of the output,

[00:31:08] Jeffrey: kind of if you, uh, if you paste a highlight from like Apple Books, you get this long Yeah. Footnote. Mm-hmm. Man, that makes me think of a couple things. One is the importance of text files becomes, uh, relevant all over again because any kind of, I would imagine people watermarking like a Word doc or something like that, all that stuff can be stripped.

[00:31:32] Jeffrey: And of course anything can be removed from a text doc, but at least if it’s all kept in good faith, you can find a way to maybe kind of bracket the, the kind of AI stuff. And like you said, to come up with like a standard footnote. Maybe you’re, you’re, you’re including what, you know, what version of the, of the model you’re using and how are you setting up that environment?

[00:31:57] Jeffrey: What instructions are you kind of coding in? [00:32:00] That makes me wonder if the best way at the moment to do this, if you can do it faithfully, is to do this work in a Git repo and maybe there’s, there’s a push that’s just, I added the AI

[00:32:14] Christina: So you had that

[00:32:15] Jeffrey: so that you can click

[00:32:17] Brett: audit trail.

[00:32:18] Jeffrey: and you can go to the, then you can go to the link just before it to be sure you’re looking at only, you know, I mean, you have to really trust the person who did that, but

[00:32:27] Christina: Yeah.

[00:32:28] Brett: sure. But the audit trail makes a lot of sense. Like

[00:32:32] Christina: I agree.

[00:32:32] Brett: we can’t expect. All of the uses of generative AI to be called out. Um, like the web is already flooded with, with generative ai, but for like in a research scenario, having a pull request and a merge his