The First 90 Days of Data Leadership: What the LinkedIn Posts Don’t Tell You with Matt Kelliher-Gibson, The Cynical Data Guy

Eric Dodds 00:06
Welcome to The Data Stack Show.

John Wessel 00:07
The Data Stack Show is a podcast where we talk about the technical, business and human challenges involved in data work.

Eric Dodds 00:13
Join our casual conversations with innovators and data professionals to learn about new data technologies and how data teams are run at top companies. Welcome back to The Data Stack Show. And today we are returning to our monthly installment of the cynical data guy, our special monthly guest, Matt Kelleher Gibson, who is jaded over decades in corporate data, and he shares that cynicism with us and add, as it adds a dash of wit for fun, as always, welcome back to the show, Matt.

Matthew Kelliher-Gibson 00:54
Welcome my time to reign over you all has come again.

John Wessel 01:01
Okay? Rain on all of us. I mean, rain on our parade. Yeah. I’m

Matthew Kelliher-Gibson 01:07
also just gonna rule with an iron fist.

Eric Dodds 01:08
Yes, the fist of data. Okay, I only have two items for the lightning round, and then we’re gonna read some funny AI stuff. If that sounds good, I actually do have a question. We’ll kind of turn the third one into a lightning round. Perfect. Okay, round one. I’ll preface it slightly. There have been a number of posts on LinkedIn where there’s a significant amount of, let’s call it sort of armchair quarterbacking. You know what you would do if you were head of data, right? Or like, you take over data in a company. So we’re not gonna, you know, we’re not gonna read any because there are so many, I’ll read a couple of samples that I pulled. Yeah, I’ll just read a couple samples that I pulled. But we won’t mention any names. Okay, so you join a company as a head of data, you gotta schedule meetings with everyone in the C suite. You got to ask the IT team for a full list of business applications. You got to create, you know, a spreadsheet from the top initiatives from each executive, etc. Let me grab another one here. You’re often pressured to deliver quickly, sometimes feeling as if you aren’t given the proper amount of time to consider data. Okay, so let’s start there. John, how do you when you see these come in on your LinkedIn feed? How does it make you feel, especially since you’ve been ahead of data? Yeah,

Matthew Kelliher-Gibson 02:35
I remember, I’m the cynical one, right? Remember? Yeah,

John Wessel 02:39
I’m gonna try really hard on this one. So I think there’s a couple scenarios. So I’ve been in like, manager team, lead role for data team, been an executive role with a data team. They had a director role. So I’ve been at a couple of different levels, all having, like, data teams report to me. And some of them have been formal, like, this is a client facing data team that we do data things and others have been like, not even like necessarily. Like, a clear boundary of like, who’s on the data team and who does other stuff. So I think there’s like, three practical scenarios of like, Hey, you start as a new data leader. One, somebody got fired. Two, they’re like, starting like they felt some need for a data team, or creating a net new team. And the third one, I think, is a variation of the next new team, is they’re like, oh, like, we kind of have some data people are gonna like, reorg and like, create a team. And I think the unfortunate like, result of any of those scenarios is it’s typically either one full of just, like, a ton of optimism, of, like, all this value they’re like, going to, like, quickly create, and, you know, save millions of dollars, or generate millions of dollars in revenue, or B the other scenario is something disastrous happened, like, but like, supporting the financials, like, erroneously for The last like, five years, like price, you know, like we priced all of our clients wrong by hundreds of 1000s of dollars. I don’t know. Something disastrous happened. Somebody

Eric Dodds 04:07
got fired. Competition is

Matthew Kelliher-Gibson 04:09
eating your lunch right

John Wessel 04:10
now. Yeah, it could be so, I don’t know that’s my initial take, because there are a couple different scenarios. And as far as how the post makes me feel, I haven’t really seen any that like, actually dive dove into what I think is a practical scenario, which is either like, almost like hysterical optimism, or like a at least year long recovery route of like stabilizing, like teams and infrastructure and like finding errors, getting, getting the, you know, punched in the gut, like, multiple times a day. Oh, like, how can this also be broken?

Eric Dodds 04:46
Do you think it’s as you read them? Do you feel like when I say these posts about, you know, if I were head of data, here’s what I do in the first 90 days. Do you feel like you can tell? Like, at. Gut level, who has actually been a head of data and who has not, not

John Wessel 05:03
necessarily, but I think I can tell that a lot of people are posting seem to be in them. You either haven’t done it or they’re just in the minority, where, because I left out the Ford scenario, which is technically possible, is, like, there’s a vacancy because, like, maybe somebody retired, and the data team was great and well taken care of, and everything was perfect. And you’re just stepping into a rock star’s shoes that had everything organized and wonderful. It’s

Eric Dodds 05:31
like, I mean, like a really nice sunset for a Supreme Court judge. Yeah, they just, they had a really long run, and, like, you know,

John Wessel 05:40
right? Or maybe not retired. Maybe they went off like, you know, they were an absolute Rockstar kill it, and they had to go start their own company. That was, I’m sure there’s some scenarios out there like that, and maybe one or two of those people have been in that situation.

Matthew Kelliher-Gibson 05:52
Have you actually met anyone in that scenario? No, definitely. I

John Wessel 05:56
have not. I do not know those. Maybe my life could be different if I did, you know,

Eric Dodds 06:03
okay, cynical data guy that was a very

Matthew Kelliher-Gibson 06:06
thoughtful, nuanced, well, well, what’s the word for it? Just a very, you know, kind take. I will say I have no room for that, so I need you to hold my beer for him. Okay? So you ask, What do I think when I see these posts? Yeah, I typically roll my eyes because they make no sense to me. First of all, because they’re always talking about and it’s a good thing. Got to meet with executives. You got to do this cool thing, that’s all fine. You know, they never talk about any of these. The team you’re inheriting, they act as though you are coming in and it’s a blank slate, and all you’re just going to go figure out all the wonderful things we’re going to do your actual first weeks. Like, that’s what you want it to be. Then the first week of you being there, you’re like, I’m really going to just sit and learn until that Tuesday on day two, when something breaks, and now they’re everyone’s yelling at you to fix this thing that you didn’t even know existed until five minutes ago. And also just, you know, you have a team. Who knows how good the team actually is? If it’s a smaller company, you probably only have a few people you need to know. Like, what are they good at? What do we actually need for things like that? They also just always have this feeling of, I’m gonna go to it and they’re gonna tell me where everything is. It never knows this. It doesn’t want to spend their time telling you this. We’re trying to figure it out. So true. They don’t want to do that. So it’s like, how well, how am I supposed to figure this out? You got to just go do the work. Like that’s what it is. They all assume this is like a startup, and it’s not. It’s almost always a turnaround. Or the worst one is it’s not really, it’s not technically a turnaround, because nobody will admit that it’s actually a turn around. So part of your job is convincing everyone who’s like you came in and they told you about all the wonderful things they’re doing. They show you this slide with their tech stack on it. They talk about how great the team is. And then you step in there and you go, this is all chewing gum and like toothpicks, and half the team doesn’t even know what they should be doing. And now you have to try to try to communicate that to people who are like, no, no. We’ve been doing great. We’ve been doing great. Well, when you know when Mike was here, everything ran fine. It’s like Mike was just making up data and lying to you. That is

Eric Dodds 08:34
such a good one. That’s like, when Mike was here, everything ran fine. And it’s like, yeah, like, how do I tell you that you’ve been looking at things like that that

John Wessel 08:44
are just, like, patently false.

Matthew Kelliher-Gibson 08:47
So when you ask the question, like, can you tell who’s done it before? I’m like, Yeah, I can tell. Because a lot of times this isn’t everybody, but I’ve looked at a couple of these posts and gone, wait a minute. Who is this person? And you see CEO in the title, and you’re like, Okay, maybe they have obviously done something, until you go to their experience and you find out they were like a developer or an analyst for like, six to 18 months, and then they started a company. They have no idea what it’s actually like doing this. And they wouldn’t. It’s not like their fault that they don’t know, but it just comes off as very well. Let me tell you how this would work. Okay, huh? Sure, that sounds wonderful. The

John Wessel 09:30
things that are one of the hilarious things to me are two, one thing that I haven’t ever seen that that is, yeah, I would say that I think a majority of people probably haven’t done it, that are posting about this, because the thing I haven’t seen that I would always start with was some sort of, like, monitoring of things breaking. Like, I never see people say that as the first step in these things, we need to implement a way to know if things are working or not working. Never see that stuff. Form,

Matthew Kelliher-Gibson 10:01
and I’ll point that out with that. The other thing is, your mean time between horrifying discovery is gonna be about five minutes. You can’t change anything when you’re like, all right, what are you doing? You do what? Okay, wait, we gotta fix that. Wait a minute. Wait, you guys pull data. How is that on a Google Sheet? If that guy doesn’t work here anymore, it’s on his account. Like, you can’t when you’re in the middle of that. Yeah, it just happens. Real stories. Oh, therefore that it does. Yeah, it totally does. And it’s like, you can’t change stuff when you’re in the middle of that. You have to kind of like, stowed away, like have a notebook where you write it. And then once that meantime between horrifying discoveries gets bigger, you can start to prioritize some of these things. So

Eric Dodds 10:53
Okay, I have another. I have a follow up question to round one. Great takes. By the way, I actually thought, you know, John’s agreeable take was delivered very diplomatically, as a chief data officer would deliver it to, you know, a variety of stakeholders, internal

John Wessel 11:12
but upon further analysis, actually, so fairly cynical was

Eric Dodds 11:15
That was pretty cynical. Actually, I’ll tell you. The cynical part was, you the best part was when you just slightly change your tone of voice and you’re like, maybe someone retired, like, man,

11:27
maybe

Eric Dodds 11:28
okay, even just if we think about the last three or four months of the show, we’ve had different people who’ve had head of data positions. And one of the interesting things about those discussions is they vary really widely, right? So, as opposed to my first 90 days as a, you know, CFO, my first 90 days as a CRO right, you know. I mean, of course, there can be variations in some of those roles, right? But the chief data officers that we’ve talked to on the show and the discussions we’ve had the role can look drastically different from company to company. And so I think another reason that it’s funny to see these posts is, I think in many ways, the role at an executive level is still being defined, right? And there’s, my sense is that there’s a lot more variation for an executive data role than there would be for something like, you know, a chief revenue officer or something. There’s

Matthew Kelliher-Gibson 12:34
also a lot of companies that aren’t hiring executive data roles. Their job description is super Tech with, you know, and do the budget or something, but really, just make them do stuff right so that I can get my report faster,

John Wessel 12:52
right? Yeah, and I don’t, and I think a lot of companies don’t necessarily need them, they, for sure, don’t need a CIO. CTO and a C date officer at a 250 person company like I think that’s the problem of there’s a group of technical roles, and you probably need, like, one technical executive at a certain size, yeah, like another size, maybe you need more than one, but when you pick I’m actually optimistic on CDO becoming more common as a technical executive, as people use more SaaS tools and, like, there’s less infrastructure, because that’s why I like CTO or more technical. Oh, like, interesting is because of the infrastructure, they had to put some money in charge, right?

Eric Dodds 13:36
So, okay, so you’re bundling your stack and paying SaaS vendors to manage it for you, and so the footprint of the technical problem becomes about integration, data integration, and sort of, yeah, fascinating, yeah, fascinating.

John Wessel 13:55
I think that’s a reason why you could say, hey, we have a CDO and not a CTO or CIO, where, in the past, like, of course, you have a CTO CIO, and then maybe you had a CDO as an add on if you needed that for your business. Yeah, yeah.

Eric Dodds 14:09
I mean, one interesting thing that I was talking about, I was talking with someone who was actually in product at a really large e-commerce company, and they do a bunch of consulting, you know, on the product side of things for E-commerce companies. And one big trend is as platforms like Shopify have become robust enough to serve really large companies, which wasn’t, I mean, that’s actually like, that’s relatively new. You have these engineering teams. There’s like, we had 20 people like, managing this intense e commerce application, and now they’re migrating that over and, you know, cutting the team by 60, 70% you know, yeah,

Matthew Kelliher-Gibson 14:49
What I do hope we see with some of that in your scenario is we’re still mostly pulling CDOs and VPs of data from the infrastructure side. Former. Engineers, for architects, things like that. And I think that is something that hampers your ability when you want to move from, hey, we just need to keep things running so we want this to be a value add. We need people who have experience closer to it, the analyst, the scientist side of it. And right now that’s just not the profile that people are hiring or Yeah, but interesting to give you one last idea on what it feels like. I remember when I went into senior director role and I was and I had lunch with with a friend of mine who was the president of a of another company, and the first thing he asked me at lunch who was my week two I was on the job was so you figured out who you need to fire yet, what

John Wessel 15:46
was your answer? No, we

Matthew Kelliher-Gibson 15:47
I did not fire anyone in that one. I managed a few people out of positions, but we didn’t end up firing it.

Eric Dodds 15:53
Yeah, that is a, really, yeah, that’s a, I think that’s, I mean, that’s the rough part of it, coming in and taking over a position. But I think in many ways, you have to go in with that lens, right, like, depending on,

John Wessel 16:06
you know, and I haven’t seen a single influence or post to include that as part of the striping over stocking, I

Eric Dodds 16:12
know, yeah. I mean, yeah, I said one of my friends held, you know, sort of marketing and product marketing roles, and then a bunch of executive roles for a bunch of Silicon Valley companies. And when he was interviewing that, I mean, he would essentially, sort of build a map if he was inheriting a team before he even took a job of like, I want a pretty good sense of like, who needs to be fired? Like, you know, etc, yeah.

John Wessel 16:35
I think another take on the executive thing that all of them mentioned, like, you need to align with the executives on things, I think that is really glossed over in a lot of these, like, posts true and that, like, I’m sorry. Like, and having been in this position, like, when’s the last time you met with an executive, they had a clear idea of what they want to do with theta. Like, never

Matthew Kelliher-Gibson 16:54
or or

Eric Dodds 16:58
The show is really taking a cynical turn here. Here’s

Matthew Kelliher-Gibson 17:01
the other version of that, because I’ve led an internal consulting group at a company where our job was to basically help you do everything better. And they made a big announcement about it being a center of excellence, all that type of stuff. And the reaction from most of the departments was essentially from the VPs, which sounds great. I’m sure other people need it, because I know how to do my job, right?

Eric Dodds 17:25
Yep. One other thing I would say is that I haven’t articulated super clearly. Let me, you know, talk with executives and understand priorities, but I would actually say another big one is understanding trust and credibility. And you really need, you really need a map of that yes, because it actually doesn’t matter if, even if you figure out what other executive priorities are, if you don’t understand how much they trust you, then it doesn’t matter. And

Matthew Kelliher-Gibson 17:55
all of those kinds of informal power structures. And yes, yep. I mean, the farther up I’ve gone and John, I’d be interested in your opinion on this. I feel like there’s a there’s like, a whole kind of map and knowledge of what you need to do at once you get above a certain level, but that it’s like secret knowledge. No one talks about it. They always give these very positive, glossed over answers. And it’s only if you can get to know one of these people and really talk to them that they’ll get into the details of, like, oh yeah, we’ll see. You got to go in there, and you’ve got to be able to fire these people, and you’ve got to make these decisions, and you got to structure how you’re doing your work this way. But like, no one wants to talk about that,

John Wessel 18:35
yeah, which comes back to the trust thing. So I think a lot of these posts get into, like, you need to align with the data priorities of your executive like is where it starts. Is like, no. You need to, A, understand all the executives’ perceptions of the problems. B, you need to go figure out what the actual problems are. C, you need to figure out alignment between the perception of the problems the actual problems, and then come up with a plan to like how data can help with the actual problems, and then align that with the perception that with the perception of actual problems, which

19:04
never matches. Yep, that’s

John Wessel 19:06
the hard part. You

Matthew Kelliher-Gibson 19:07
also need to figure out which executives do I need to be willing to do favors for that are things I don’t want to do versus the ones that they’re going to always ask for this I’m never going to get anything out of it, so I’m not going to do this for them, because you will get stuck in that. You know, it’s like the 8020 rule. They’re the people that will spend 80% of your time. They will completely block you up, and you will get nothing for it, versus the people that you should be like, Hey, I know if I give a favor, and if I do this, this guy over here, he’s part of this informal power structure that will give me leverage to go do other things. I

Eric Dodds 19:43
I also love that we heard about the art of war over here. We just, we just called on everyone who’s been making these posts, and then we just said, like, here’s, you know, yeah, so we’re guilty as charged. I think I might have to make a post now, but yeah, you probably will. Okay, that was a great lightning round in. Insightful, healthy amounts of cynicism from both the cynical data guy and the agreeable data guy. That was a good one. I’m proud of that. Okay, that’s a new standard for me. All right. Lightning round number two. And then we’ll go to AI fun. I’m gonna read this one out. Okay, this is interesting. I’ll read the whole thing. And this is Sebastian Huey. We’ll give you a call out. Sebastian, reach out to us on the show. We’d love to have you on if you would like to join every thought provoking post. There’s no difference between head of data and head of product. Both must build products that make the company money. The operating model doesn’t differ between a head of data and a head of product. And this is a list. Talk to users all the time. Identify user problems and hidden desires, understand users’ jobs to be done. Figure out which problems are really painful, build MVPs to address these problems. Measure user adoption using qualitative and quantitative feedback, iterate the solution until it reaches product market fit. Start again from the beginning. It took me years to figure this out, but once seen, it’s impossible to unsee data. Teams that struggle are usually ones who haven’t seen it or don’t want to see it. There is no difference between head of data and head of product. All right, cynical data guy,

Matthew Kelliher-Gibson 21:10
I don’t know enough to be able to say if it’s the exact same as a head of product, but I definitely think having a product mindset will help you. Definitely that is one that I think one of the things that had helped me in my career was I embraced more of that idea. This is a product. You got to think about it. It doesn’t matter what I think is best. I got to go talk to the people who are going to use this and figure out what it’s going to do. I mean, even one of the best kinds of insights I had was looking at it and saying, there’s kind of two ways you can build stuff for people. One is that we built this thing. It’s amazing and it’s fantastic. And if you would just do it the way we tell you to, yeah, it’ll work great. The other one is, hey, you do this thing. It is painful for you, I built it into your flow, right? And that makes a big difference on just, will it get adopted or not so, but, I mean, I will defer to the actual head of product in the room, yeah,

John Wessel 22:14
oh yeah, Eric’s take. Oh, wait,

Eric Dodds 22:16
So am I going next, not you?

John Wessel 22:18
Yeah, okay, you’re the head of product.

Eric Dodds 22:20
I think it’s really valid. And I think I would reiterate what Matt said in that it’s really about having that mindset on building anything right. I mean, that’s a helpful mindset, sort of no matter what you’re doing. And I would also say, I think one thing I like about the post is that it is the list that he writes out for, okay, how do you build a product? It’s pretty long, and that’s really important, because if you miss any of those steps, it can create severe problems, right? But the reality is that in a lot of product organizations and data organizations, whatever, you’ll skip a couple of those steps, right, right? And I think that can be really problematic. I think where I would disagree on some level is, I mean, probably, I guess if I met Sebastian, we’d probably be, we’d probably be very aligned. The first sentence in the post, I think, is just overstated, but that’s LinkedIn like there’s no difference. I don’t agree with that. And one of the big things, there are a number of things, but I’ll tell you the biggest, one of the biggest things that is that came to mind immediately when I read the post was, and we’ve talked about this before, actually, we talked about it recently, where we talked about, it’s okay to build something, you know, build a data product for two months, and then if you get 80% of the answer, you just throw everything away. That’s like, that’s okay, like, you should afford the exploratory nature of data. Yeah, the ability to store immense amounts of data forever has just never been easier or cheaper, right? Of course, technical debt exists, but it is way harder to kill a product that end customers are paying you for, yeah, that is like, deeply tied into, you know, your software platform. That’s hard, you know, for a number of reasons, and you cannot deprecate things. You just can’t deprecate things as easily, you know. And I think that you should be willing to do that if somebody’s not worried, like, of course, you know, those are sort of core principles there. I don’t want to describe that as saying, like, the stakes are different, because I think that, you know, it’s just that the consequences for making certain decisions when you roll a feature out to production, you know, and people start paying you for it is, you know, that’s a different dynamic, at least in my opinion. John,

John Wessel 24:53
yep, so no, I actually know Sebastian, he’s great. Runs that action mentor, and I agree. Of philosophy, for sure, they like product philosophy here. I think the trouble is, some data teams aren’t set up to

Matthew Kelliher-Gibson 25:08
be able to do that. Some also don’t want to.

John Wessel 25:12
That’s true. That’s true. But if I were to give it like, let’s go back to the head of data thing, if I were to give advice to a head of data thing, it would be absolutely to have this mindset, and then the practical application of this mindset is exactly how you ended Eric, number one, you like associate your data team with some revenue generating activity, yeah, or at least with some like forward client facing activity, yep. So where it’s like, are you producing analytics that clients see that’s good, even better than that? Are you producing, you know, some sort of data analytics that somehow impact revenue? Because if you’re just stuck in like, we only produce accounting and operational things like that, that is an easy target for, like, oh, we need to lay off some people, yeah. Like, it’s an easy target, yeah.

Matthew Kelliher-Gibson 26:03
But only caution I would give to that is I have worked with people where, if you don’t understand the situation, or like, they don’t want you to be a product person, it can get you in trouble. Sure. I mean, I worked with a guy who was in charge of BI, and he came in with this idea of building a platform, like an internal platform, for people to self service, you know, self service. And he viewed it in a way which was like, we’re building a product. So he was primarily brought in, like B by developers, and when we got into it, and when, you know, and by the end of his tenure there, what became really obvious through it was he didn’t want a platform. Yeah, they want you to know that he wanted reports, which is really what they wanted. And so he had, and I don’t blame him for doing this like he he had been sold a certain vision and was going on, and it was only when he got, you know, six, 812, months down the road that it became super obvious, yep, because, as I said, they lied to you about this stuff, right? I mean, he had, he built a team that was completely wrong for what the need was, yep,

Eric Dodds 27:21
I think the parallel math that you drew with, I mean, both of you getting close to the problem. I actually think one of the interesting things that I would add on to Sebastian’s post is thinking about your data like the individuals on your data team. Like, are they actual product managers? Right? Like someone who is just, I have like, raw skill as an analyst isn’t a product manager in the sense of, like a product team, necessarily. It can be, and the best ones often are, but I would say that’s like a great that that would be a huge way to actually achieve running a data team like a product team. The hard part is finding people who can do both of those very difficult because very especially if you’re someone who came up with that mindset, yeah, you because I did this, you overestimate the people who want that mindset and that they can develop it quickly. Yeah. I’ll give you one example, you know, of understanding, you know, building season, some in some downstream problem is attribution. Attribution reporting is something a lot of companies struggle to build, and it’s one of the reasons is that because running advanced digital marketing is extremely complicated, and understanding how the platforms deliver ads and the way that data is captured, like, it’s pretty like, you know, you know, you have to have a pretty good knowledge of that, how to use your, you know, all that sort of stuff, right? And so, which is incredibly important for understanding, you know, how to shape attribution. Okay, we are

John Wessel 28:56
one other quick, quick thought. As far as I guess, I’m full of advice today on the data team thing. I think the other thing to think about where, like, Matt was talking about this too, is it’s, it’s so easy to get stuck on, like, the current, like, what is, like, what the data team was doing, and it is really hard to break out of that, and then to come in with a vision of, like, we’re gonna change everything. We’re gonna change everything. We’re gonna change the culture. We’re gonna change all the people. We’re gonna be doing something completely different like that is very hard, and I think sets most people up for failure. And I think a more reasonable approach, like your friend that, like, did the platform thing self service, would be to come in and, like, go for things like stability and trustworthy data and accuracy, and then like, build off of that into something versus, like, coming in, like, Yeah, we’re gonna change the world, change everything and

Matthew Kelliher-Gibson 29:46
rest. And the hard part with that is, is that you have to be thinking in years. Yeah. You have to be thinking for years. And the average, like, CDO, 10 years, like, two years, yeah. So it looks a little bit like being a football coach in that sense. Some right? You have to set up the right foundation and the right culture to be successful on the long front, but also when, if you don’t win the next two years, you’re going to be fired. Yeah, okay,

Eric Dodds 30:10
We are really close to the buzzer here. So super quickly. Have you seen the AI summaries that Apple is delivering? So there’s a great CNIT article. There’s probably a bunch of these, but this is from the article. The other day, I received several text messages about how bad a hike was and how that person felt dead tired, along with some other sparse details and Apple intelligence summarized those messages for me, hike extremely difficult, almost another one. This is a text I got about how a trainer killed my friend with a particularly tough workout, Apple. In summary, the trainer allegedly killed needs to go to hospital. Okay, here’s my hot take, but I want y’all quick hot takes on this. This is funny, but I also think that it makes it hard to appreciate, like, the actual value of AI sometimes, or like, to control perceptions around the technology, yeah, like it’s, I hope it’s not damaging. I hope it’s not damaging, right?

Matthew Kelliher-Gibson 31:15
Well, I mean, the early days of Siri had some similar stuff, and there’s a whole bunch of people who just never used it, never wanted it, because it messed up early on. Yeah? So, I mean, it’s a tough one, yeah,

Eric Dodds 31:26
yeah.

John Wessel 31:27
I mean, it’s an interesting take, actually, from a company that’s very cautious about releasing early totally. So that’s the fascinating part to me. I had another one on here that is a one from Red, from Nick for anyone who’s wondered what an apple intelligent summary of a breakup text looks like. No longer in a relationship wants belongings from the apartment, and he basically goes on to say, like, yeah, this was a good summary. And there’s that it’s but there’s that tone thing and that like, like Nuance thing, yeah, that’s clearly like, they haven’t quite mastered yet. Yeah.

Matthew Kelliher-Gibson 32:02
I mean, you see it on, like, Slack, AI too . I was just reading one today where it said that, you know, John Smith made a funny comment. That’s a weird summary. And I clicked on it, and it was literally a post that said funny comments and a link. Yeah, the person didn’t make the funny comment or whatever so but I mean, like, you said, like, it’s, it’s getting it right in some ways. And this is, you know, still impressive for a computer to be doing that. Yeah, totally. Tone and understanding, yeah, sarcasm, not their metaphors, not there, completely, yeah,

Eric Dodds 32:38
yeah, the many facets of AI. All right, we got to close it down. If you have any funny AI summaries, send them our way. You can go to the website, fill out a form, or send us an email, and we’ll talk about them on the next show.

Matthew Kelliher-Gibson 32:52
Ooh, yeah, yeah, stay cynical,

Eric Dodds 32:56
catch you on the next one. Yeah. The Data Stack Show is brought to you by RudderStack, the warehouse native customer data platform. RudderStack is purpose built to help data teams turn customer data into competitive advantage. Learn more at rudderstack.com.