Skip to content
S4E8  Will We Ever Love Using Software? episode artwork

Sep 12, 2024 1 hr 4 min

S4E8 Will We Ever Love Using Software?

from Invisible Machines podcast by UX Magazine Hosted by Invisible Machines

As AI agents continue to alter the way people interface with technology, Robb and Josh were eager to speak with journalist Matt Alston about his popular Business Insider article “Everyone hates Workday.” In this deep exploration of our multifaceted interactions with machines, Matt reminds us that Workday’s UX is different across a whole spectrum of users, and that generally speaking, people don’t like using software. The patterns we follow when operating software end up shaping our thinking, and as we move into a world dominated by conversational interfaces, it’s important to consider the way these systems are intertwined with our minds. Don’t miss this eye-opening episode about the future of project management. Learn more about creating and orchestrating AI agents to streamline your business operations: https://onereach.ai/ai-agents/?utm_source=soundcloud&utm_medium=social&utm_campaign=matt_alston_episode&utm_content=1 Article, “Everybody hates Workday”: businessinsider.com/everyone-hates-workday-human-resources-customer-service-software-fortune-500-2024-5 Learn more about Matt Alston: mattalston.com. #AIPodcast #InvisibleMachines #AI #ArtificialIntelligence #ConversationalAI #AIAgents #UXDesign #Workday #ProjectManagement #FutureOfWork #UserExperience #SoftwareDevelopment #TechInsights

Episode details come from the podcast's RSS feed.

Embed this episode

Ready to play

S4E8 Will We Ever Love Using Software?

0:00 1:04:12
of matches

Transcript Auto-generated, may contain errors

Welcome back to Invisible Machines, a podcast where we explore the expanding world of AI agents and conversational technologies. I'm Josh Tyson, co-author of Age of Invisible Machines, a bestselling book all about the agentic approach to automation and the broader pursuit of organizational AGI. I am joined each week by Rob Wilson, CEO and co-founder of OneReach.ai. Rob and I wrote Age of Invisible Machines over countless hours of conversation, much like the ones we have each week on this podcast. Rob has spent decades researching and deploying conversational technologies, and Invisible Machines is where we share big ideas with fellow AI insiders. We're continuing these conversations in partnership with UX Magazine because attention to experience design is absolutely critical to the proper deployments of AI agents, or invisible machines, if you will. So with these AI agents continuing to alter the ways people interface with workplace software, Rob and I were really excited to speak with journalist Matt Alston about his popular Business Insider article, Everyone Hates Workday. Matt, who also wrote the popular 2023 Wired article,

your project management software can't save you, joins us for a deep exploration of our multifaceted interactions with machines. Among other things, we look at how the UX associated with productivity software like Workday is distributed across a whole spectrum of users who have their own individual experiences with it. Generally speaking, however, people don't seem to like using software very much, which is noteworthy because the patterns we follow when operating software end up shaping our thinking. So as we move into a world dominated by conversational interfaces, the intertwined relationship between our minds and these tools matters more and more every day. This fascinating conversation with Matt draws from his own experiences using software like Workday as a well-established copywriter, opening a rich discussion about the work of getting work done. So let's get started.

Invisible Machines is produced for UX Magazine in partnership with OneReach.ai. Named a leader by every major analyst group, OneReach.ai's generative Studio X platform was built specifically to facilitate the no-code orchestration of AI agents. GSX is trusted by global giants like PepsiCo, PwC, Deloitte, and DHL, but anyone can experience the level of agent orchestration we talk about on this podcast through interactive demos at OneReach.ai. Now, on with the show. Yeah, Matt, so it's interesting. I feel like it's kind of really difficult to make an experience that bad even worse. So there's almost like a weird feat there. But it was interesting because we had a conversation recently with Greg Vert, who he leads digital transformation for Deloitte. So he's working directly with HR departments at large enterprises, helping them out with this stuff. And he pointed out that Workday, you know, actually kind of gained traction initially because they replaced whatever kind of archaic HR system or software was available then. And they were renowned for their experience.

so that kind of opened up this interesting hole in our brains like you know what does it mean when a company that was once really good at ux has become uh emblematic of how bad you can be at it the thing that i sort of kept circling and i think is like ends up being the thesis of the piece is that um who's the user like who's like a good user experience for whom um if i'm an hr person at a bank of america and i the ratio of people that i serve uh to me is like 280 to one or something like that i'm making that figure out but you know there's 130 million workers at companies in the country and there's like 1 million hr professionals so like you know take out all the small mom and pops that have three or four boys and somewhere in there like two or three hundred people teach hr person like And I'm the buyer of the software. I'm the person who implements it. Like my user experience is probably very good because my binary output is, did I get somebody hired? And did it give me enough people to put future employees through that hiring funnel to hiring managers and get the answer I want pretty quickly? Like probably a good experience. I mean, I've talked to a lot of hiring managers who are also like, well, you know, onboarding a freelancer is pure hell. But I think like...

From the perspective of your question, you know, Workday is renowned for X and now they suck at X. It's probably like a shift in focus. Like probably like launching a product is all about the demo and who sees it and what it feels like for everybody. And then the longer... Entropy has the ability to set in. Lots of different people who work there have different KPIs for their jobs. They're like, well, who cares if a Bank of America applicant has to create a whole new profile to apply for a job at Spotify? I only work at Bank of America. I do not care. And that's not unreasonable. My neighbor's brother is an HR guy who read the piece. I was like, your friend doesn't get it at all. And he forwarded me a long text spiel from his brother that was like, my life is easier because of Workday and your friend Matt is a moron. And I was like, well.

Sure. Yeah, but mine is worse. I don't know what to tell you. Right. Yeah. And I think that's, yeah, I would think you could sit in a conversation with someone who says it used to be good and say it still is. And you could sit in a conversation and say it used to be good and say it's not, you know. Yeah. I think there's a cycle. I think the way to, and what interested me when I read this article that you wrote, which I, I thought you hit on something.

really timely and profound, even though it's been there the whole time. The first principle of like, does anyone like any software? We just don't ask that question because there's no alternative interface. You know, we're just like, well, some software is better than no software. And that's what your neighbor essentially said to you. Yeah. He said, well, like, the software I know, no matter how hard it was to learn it, is better than any other software out there that I'd have to learn.

because learning is painful, but not only is learning painful, it's that moment where someone comes in and says, hey, I'm going to train you on how to use Workday. And you're like, okay. Then they train you and then you know it. And then you go, wow, this is really lame. Why did they do it this way? Why did they do it that way? Why is this so confusing? And everyone goes, I don't know. It's just what they did. And then you get through training and then you get to your next job interview and they go, okay, it's between you and somebody else. Do you know Workday? And you're like, yes, I know Workday. You get the job. And now you're like,

oh, I like Workday. And I like that it's hard to figure out how to use because now I can get jobs that other people can't get because I went through the pain of learning it. So now I like Workday and I'm going to write long emails to people who are anti, like I'm now, like my incentive has now gone to. So like, and I like that it's hard to figure out because then, because if it's easy, then no one's going to ask like, do you know Workday? Yeah. It's not going to be part of the interview criteria and that investment isn't going to mean anything. And so I think there's like the world of people who will defend antiquated old UIs because they're incentivized to do so because they learned it and they feel like they have a skill. Like Oracle always put in this category, like it was so hard to figure out as a database, but once you knew it, you were in this club and that club was like this elite, high paid, you got paid more than everybody else. It was hard to figure out. So you had to take.

all of these classes and it was a bit of a weeding out process right it's sort of sort of like why people get jobs with a college degree from harvard because there was like this weeding out process that happened of like are you willing to put up with all the that you'd have to to get here yes okay then you'll put up with all the in my company that i give you um without quitting in my line of work as a senior copywriter like now you have to know figma like that's just if you have a team that uses figma like you might as well you know Right. You might as well be using a legal pad and a pencil. Right. And that has happened in only a few years. Right. And what's amazing, actually, if you think about it, isn't like, why do we use shitty software that's hard to figure out? But why does anyone ever change? When Workday came out, it's funny because one of my first companies was one of the first UX firms and Workday was a client in the early days when they were disrupting. And they disrupted based on UX. That's how they disrupted the SAPs and everybody else.

And you wonder, why did anyone switch? If basically the software you know is the easiest software to use, and it's a chicken or the egg, and these software companies eventually get to the tipping point of most people have already learned their software that need to know it, why does anyone even give other software an opportunity in the market? And maybe is it just because... Because these companies at some level intrinsically culturally know that they have this advantage that they just stop investing in UX and it just atrophies to the point where they're like inevitably going to get so bad that people can't ignore it anymore. Or does somebody just come out with something that just seems so easy to use?

They're like, oh, this doesn't look like it's going to be hard to learn. As an often applicant to smaller startups, sub-hundred employees, employer brand really matters. You kind of are like, wait, what is this company? What do they do? Does this seem like real people? Does this seem like something that will exist in three years? A company has to protect that. A company makes their products and they have their marketing brand, but they also have their employer brand and they need to...

you know, like to attract talent, they actually need to seem like, you know, founded by a former Microsoft or founded by a former Google, blah, blah, blah. Like, you'll see those as part of the pitch bios. And those aren't, those aren't just for Silicon Valley. Those are for like, to peel off Google or, you know, FANG guys who might, who they might like. It's the glass ceiling, right? And, but if, you know, if you're Spotify, sorry to say Spotify or whatever, I'm fine with Spotify. If you're Spotify, you don't need to protect your employer brand with a good or bad application experience. It just doesn't matter. So you have HR people who say,

Every time I post a job to LinkedIn, I get a thousand applicants. How do we resolve this? Oh, Workday can make it harder to apply. By accident or by intention. That's the purpose of a system is what it does. Yeah, we're going to go on accident because I doubt the sales pitch is, hey, if you use Workday, fewer applicants will apply. Your employees will submit fewer expense reports because of the pain of figuring out how to do it. So the user license will save you money. The ROI is in how hard it is to use. Yeah. Just to go back a little bit, I was reminded of, there's this New Yorker writer, John McPhee.

like legendary, you know, nonfiction writer. Um, like about a decade ago, he wrote a piece about the software that he used to research and write pieces. And it was like, I'm now I'm, I will paraphrase and mischaracterize the specifics, but he essentially just had like a clunky old DOS computer and some program that some guy in the Midwest or somewhere had created like bootstrap. And he used it and bootstrapped it further. And he had written hundreds of New Yorker pieces and like dozens of books just on this like computer and increasingly over the life of his career.

like he would go back and check and like he was one of the only like four or five people alive still using that software that was like still on a three and a half inch floppy that if like you know you know near a magnet he would lose the backup copy and um you know and ends up like getting in touch with the guy and the guy's like i don't even like you're still using it what is wrong with you and he's like if i change the way i write my pieces with my software it's like changing how i think and i think like sometime you know the the project management piece and inside i kind of had when talking to people was that like certain heights of UI, dragging a thing from left to right, like on a Kanban board. Like when they become ubiquitous, they become the way people actually think about performing tasks. Like, I'm going to move this from the to-do column to the doing column to the done column. Right, which means it can exist in multiple places. Like we've suddenly taken out the idea that the reality of the world, which is it can be in all three columns at the same time. I did a project a long time ago. Not that long, but long for most people.

It was all about reinventing education for a couple of the school systems, public school systems. And one of the biggest ideas was to just remove the concept of grade one, grade two, grade three, grade four. To make it more great. Like math, you might be in grade one. Reading, you might be in grade two. So it was breaking down this concept that you just, it's like the stair-stepping of first grade, second grade, third grade.

And it was amazing. It just turned it all into badges. Like you just got badges for certain things. It was a fantastic concept. It was incredible, the results. And what also happened is like the community could get involved. So you broke these education concepts down into really small pieces so that, you know, bees, understanding bees was a badge and wasn't attached to grade one or grade two or grade three. It was just understanding bees. And within boundaries of a menu system, students could be like, that's what I want to do next. I want to learn about bees. But the local beekeeper could actually like,

participate he knows what you need to learn it's published with what you need to get out of it so on saturday your parents could take you to the local beekeeper he could teach you those things and then the teacher's job was just to test to make sure you learned it right and then you get the badge right and so classrooms anyway bottom line is the results were irrefutable more like uh there was a high school i don't remember exactly where it was and that's more about my memory um and And it was like, they didn't have a single kid that had ever gone from high school to college in the history of the high school. It was a small high school in like the middle of nowhere. And all of a sudden they implemented the system and half the kids went to college. It was like incredible, like undeniable. Wow. But you tried to take it to the public school system, which was the job I was hired. Like, how could we implement this? And they're like, all of our software has like, the first thing you have to enter is what grade is your child in? Like none of the software you could use. And so none of the schools could implement this because even down to.

setting up the classrooms and the layout of the school facility itself. It was, this is the grade one classroom and this is the grade, everything required this thing. So our software, it's the whole like, I'm learning it wasn't, we shape our tools and then our tools shape us. This idea that, yeah, we created project management software and then the software that we created shapes how we complete projects and think about projects. Sorry, I went off on a complete tangent on you there, but you got me all excited about this because I feel like if we're all using the same software, then then how are we going to evolve and be better than each other? How are you going to get better than average results if we all do things the exact same way? So the idea that we have this cookie cutter rigid software interfaces that tell us how to think, tell us how to do things, how are you supposed to hire better than everybody else if you're using the same shitty processes that everybody else are using? Or maybe it's not shitty, it's just shitty at the moment everybody's doing it because it becomes average. And if you're trying to be above average, once you're average, you're shitty. Yeah. Well, the other thing about...

Like, Workday is that it's not just a hiring platform. It's also a talent onboarding platform. It's also can be, you know, like these big HR platforms can also do payroll. They can also do benefits administration. They can also do career stuff, you know, ongoing progress, check, you know, whatever, reviews. And like, those are all set of things that you have to solve for. And, you know, like some of those come out of workplaces, right?

I onboard you, I have to take your I-9 or, you know, W-4 and put it somewhere and hold onto it for the end of the year and make sure I scanned your passport or whatever. And like, that's easy enough, right? Like that is a problem of scale. One person doesn't want to do that. What if we just make everyone enter their own and we just check it? What if we scale that software? Great. Like easy. Like that's great. Same with payroll, but like.

As soon as you introduce complexity, you have to have humans involved again and a software just like can't start solving. Right. It's rigid. Yeah. And then you get workarounds. Like how do we put this into three buckets? Like it's, like you said, that example you gave, you know, how do we put it into three buckets when the software doesn't let it exist in three places? Oh, I got to create three instances of it. I'll call them three different tasks, even though it's the same task. And you're like working around it and then it's really confusing, right? Because you're like, well, why is there three tasks? Oh, because I couldn't put them into.

The same swim lane. Yeah. Or to speak about product management software, like my job as a copywriter, I have two or three ongoing projects that might take me a few weeks right now. Uh, one or two quick hits. And then, you know, someone slacks me and says, Hey, can you look at this? Can you copy? Can you proofread this real quick? My, I do it. I send it back. My, my.

project management software treated brain like you have the impulse almost like as a former bartender I'm just like always want to wipe the surface down in the house like yeah it's like my brain is like should I log adding that apostrophe to that guy's like thing for the you know push notification put it on the board drag it to the right and call it done like did anyone see my work did it happen like doesn't especially remote especially like more digital workforces like you know a lot of these the Kanban the left to right or the Gantt chart the top to bottom dependent you know x can't happen until You know, A can't happen till B happens till C happens. Those are from like building the Hoover Dam. Those are from real roads moving west. And like, I work with a designer. Our processes aren't necessarily dependent on each other. She can be, you know, laying out a website and say, copy TBD, or she can say, I can't design this to you. Give me the copy. In either case, like...

We're almost interdependent systems and a structure that either links us as like consequent left to right or like dependent waterfall, like can't do X until Y. Neither one of those really like lines up with some of the swirl of like modern work, especially iteration, especially ideation. You know, it's easy for a product team to say, you know, we want to improve conversion on X, but what are the variables? How are they dependent? Are they?

you know, as soon as you have three or four people, the combinations, the sort of factorial combinations of how many different ways something could be done don't really fit a system that is designed to be simple, I guess. Right. Yeah, I think when you look at software itself and you go to first principles, you say, do people like using software? And the answer is yes, if it's a game. No, if it's productivity. Why? Because people don't like to do boring tasks. But we know with gaming, they really enjoy software. They get addicted. It could be addictive. So software can be anywhere from hate to love it so much you can't put it down.

And then you sort of ask yourself, well, why is most of the software that companies use so hated? If it could be loved, could it be loved? And the answer is probably because it's not designed to be loved. It's designed to make you more productive, theoretically. And then there's a big question mark on does it make you more productive or not, to your point, or does it just make you more transparent and less productive?

Like now, now somebody can see what I've done and feels like I'm busy, but am I more busy than I was or am I less busy? Cause I'm so busy filling out the software. Yeah. To make sure everybody knows I'm busy. To the right convention, putting it in the right folder. Yeah. Like the transparency part is, is, you know, also, uh, an important distinction and different about like, cause again, all these systems came out of factories and construction and, um, you know.

really like hierarchical systems where a person sat up in a balcony and looked down and it's like, that guy needs a break. Or, you know, that one is not as, that guy's not as productive as that guy at building lathes or whatever. And so oversight is also like a huge factor in like how these systems work for the people who aren't the individual contributors, but who are, you know, the bosses. Yeah. I think that if you go again, like I hate to keep rehashing first principles, but I think if you go up to the fundamental fact that most people don't like the software they're using, period.

And most people dread learning new software. You go, software itself has a fundamental issue that we're not addressing because we haven't seen an alternative. And so you go, well, what's the fundamental issue? It's probably what you said, which is one experience, one cookie cutter experience that's supposed to serve a large group of people. And so even if the software is configurable, which the bigger the company gets, the bigger the company that uses your software.

In other words, as enterprises start buying your software, you have to make it more and more customizable. And that creates all of this complexity. But, but even still, it's that one size fits all. And the fundamental problem is our software is not adaptable enough. And then you see AI and you see this new interface.

about humans, which are very adaptable. Like we know this because every call center puts a human there. Like that's the most adaptable user interface there is. And we go, okay. And then you say, what if humans could conjure up micro UIs, like little UI components and pictures and songs and things like images and videos on the fly to even improve their communication. So like now we go to cookie cutter, to hyper personal.

the other end of the spectrum. Now you can have a hyper-personalized conversation with a machine where the interface is absolutely designed for you because it has your context. Now these companies don't have a stronghold on the sort of cost of learning something new, right? And it's like, there is no learning anymore. I just ask the software, like, you know, hey, I want to, you know, can I submit this as an expense? It's like, sure, I just did it for you. Yep, yep, yep.

And you're like, oh, well, thanks. Now, what system are you using? Who cares? Like, wherever that data goes, it doesn't matter anymore. It just became, you know, as the book we wrote says, Invisible Machine, it became invisible what software you're using. The machine became invisible because you don't care if you're talking to a person or a machine. And so it's why we think that this change, this, you know, change from traditional UIs to this new micro hybrid UI language is going to be so fast and rapid.

massive is because software UIs as they exist today fundamentally suck. Yeah. I have a product guy that I work with who's training in a GPT to write his PRDs. Okay. You are this. You work for me. You know, as everyone he makes himself, he puts in and it's increasingly giving him much better and more specific PRDs for all the different kickoffs that he wants to do. And I think you're right. Like, his boss surely doesn't care what the process was to get the tool to kick off a...

sprint to get the result that we wanted a company. Do you think, and like, I won't care as a consumer someday if instead of Googling whether in Portland, Maine this morning, I just ask, I already do ask my little hockey puck on the counter, but like, I don't care what the gear turning on the other end is if it increasingly gives me good results. Do you, could you imagine that that same thing for HR tasks would reach and pass the tipping point where someone doesn't care? Do you think like someone was like, no, no, I actually want to know where the form is. I need to see it. I need to see. I need to see. Cause I don't believe that the tax return owes you $3,000 in expenses or whatever. Right. Yeah. Look at the tax return. Like it didn't, it didn't take long for us to go. Like, we don't want to fill out the 1040. Just ask us a bunch of questions. Like Intuit got that super early. We don't want to fill out the 1040. I don't even care if I see the 1040.

Yeah, that's a good thing. I don't even want to know what a 1040 is. I just, like, ask me a bunch of questions, I'll answer it. The answer it was, like, way ahead of the game on conversational EYs. It's just, why didn't everything go in that direction?

Because I think, almost like oil, you know, I think there's a lot of money right now in traditional UIs. And I don't think it's conscious to your point. I don't think it's that the buyers know. I don't think that the sellers are really aware of it. I think the fact that UX is always this sort of side show in every company, you know, except Apple maybe. Because it's always relegated and siloed and like, oh, this thing we have. Oh yeah, make it look good.

I don't think they realize that most of the intrinsic value, that app itself isn't an app without a UI. It's the bundling of features that make it an app. You take away the app, make it a bunch of API calls. It's not an app anymore. The app doesn't exist as a thing. The only thing that makes it an app is the bundling of features together. The only thing that bundles them together pretty much is the UI. And so now if an app is a UI and we've changed UIs...

and we get rid of them, what is a company, what is an app company anymore? What is Workday without its UI? And it's just a bunch of API calls. That's all Trello and Asana are is, you know, API calls that look like a database. Right. And we see this with these companies that are going, okay, well, we're going to put a conversational UI in sidecar. Like we're going to sidecar the traditional UI with a conversational UI. But then they immediately start doing what you're saying. Oh, but wait, why wouldn't they want to do all of their tasks here? So they start to like sort of gobble up the world of software going,

all these companies now competing with other software vendors that they never thought they'd compete with. Workday competing with, you know, accounting software. And so at OneReach, we sort of call it the Great Convergence. And my colleague, Kevin, calls it like Battle Royale is coming because all of these major, major software vendors are all converging and suddenly going to compete with each other where they never saw them in the space. It's like Gartner is sitting there going, all these magic quadrants are turning into one magic quadrant. You can see the battle beginning at Gartner for who's going to cover what because they're all converging on the same space because the UI is vanishing. And I guess the question is, how long can these companies, just like the oil companies, how long can they hold on to this until the market says...

you know, why would I keep paying for these traditional UIs? The concept around this is point solutions versus systemic change and that we're at that point where Workday adds AI and that makes it more valuable. And everyone's like, wow, okay, I'm going to invest in Workday because they're going to bolt on these AI add-ons, right? They're going to improve the shitty UI experience with point solution bolt-on features that are AI driven. Summarize this, summarize that. And all of a sudden, it's going to fool you. Right now, yeah. Yeah. Oh, wow, they're starting to make more money because of these AI bolt-ons. But then systemic change is like right behind it, which is the removal of the UI altogether, the traditional UI altogether.

which is right behind it so you see this like climb of like oh maybe they are going to survive and then all of a sudden bam the disappearance of the app happens and and you go like am I going to use Workday to get all my work done? Is Asana and Workday going to... This is where I was going with this long, long rant. Is Asana and Workday going to join together? And when were they going to join together so that you're just going to use both of them with the UI? And then what is Asana and what is Workday after that to your neighbor who's like adamantly defending Workday to you in that long email? Yeah, or like when does it hop the fence? And when does like Manpower or Adeko say, now we're the...

We aren't the people who put people in touch with other people for jobs. We aren't the massive 30,000-person hiring company. We are the hiring company and platform. You could go in either direction. You could go from Workday toward other UI, or you could leave Workday, which is UI, and go toward the actual task, which is hiring people. Right. You're like, yeah, because a person just says, hey, I want a job at this company, and they say that to their AI assistant. That's it. Nobody's using anything.

Yeah, it seems like there's like a moment too where, you know, these companies, like if there was a big company that was really, say, fed up with the experience our employees were having with Workday, you know, there are tools out there that enable companies to start building their own ecosystem where they have a conversational interface that obscures all the messiness of Workday and also like other software that you think starts orchestrating it. And it seems like in a long enough timeline, that actually, we read about in the book a bit, but if you pursue that goal, you're essentially starting to build a digital twin of your organization. And the more robust that digital twin becomes, it actually... I think would start kind of mirroring that software. And so then that software becomes irrelevant. It moves so far to the background that now the ecosystem that you built to try and manage all this disparate software has now kind of made that software useless because it's adopted all the things it needs from it into its own password. It's this idea that our AI, sorry, Josh, I'm getting excited here. The AI is going to operate our old software. That's the bottom line. Our current, we know it can. Like that's, now we know that LLMs can operate our old software, especially when it comes to using APIs. But even when it comes to clicking and pointing and clicking.

We see this with Rabbit and some of these other technologies, RPA, creating RPA scripts, blah, blah, blah. So if AI can operate our old software, we're like, oh, cool. Now I don't have to operate Workday. It will submit my resume for me using Workday. Cool. Workday's happy because they're still in the mix. And now there's an AI company that's happy because they're operating Workday. Except at what point...

Does any, anyone, any engineer, anybody with logic step back and say, wait, why do I need Workday? Like, because you start to realize that Workday becomes a hindrance to the AI. In other words, to your point, it's so rigid that the AI is going like, we don't need this antiquated thing in the mix. Let's skip the part where I put it in Workday. Then George goes into Workday to see what's been submitted. And I'm just going to send him a Slack message that says, hey, this has been submitted. FYI, skip the Workday piece.

and say, well, the only thing then is like, where are we going to store the data? That's all. It's just a data storage problem at that point. And you're like, well, well, you know what? If AI can operate the software, it can also retrieve data and put data and store data.

And maybe I don't need to know where the data is stored because it just needs to know where it's stored. Yep. Which we know it can happen. So it's just going to store it wherever. I don't care where it stores it as long as it knows where to get it. I already don't care where it stores it. Like, anyway. Yeah. Just to your point, like right now you enter something in, your buddy, right? It goes into memory. Like, does he care where? He doesn't care. Where? He's not like, oh, where's that memory? Is that in Workday? Is that? He doesn't care. He just cares that the system knows where to put it when he wants to retrieve it. Yeah, exactly.

And that is safe. Yeah. I met with this grad student, David in Berlin. I interviewed him around the project management thing. He was a design student who was, his whole thesis was like the graphic and visual design of project management systems. And he was super interested in like the, how it looks and the, you know, your tool shape, tool shape, your thought of it. He was like, you know, not, not like blue and green, but like, you know, the, the sort of.

Windows versus databases versus spreadsheetification of these different things. And yeah, he was like, I think the end result of this is that none of this is part of it anymore. You just say, what am I working on today? Like, I told, you know, these are the projects. What am I working on today? And then he says, you're working on this. It functions like a de facto person project manager, you know, because I also think like interviewing a lot of people and just having a ton of experience with the project management stuff.

The best systems are the one where one person is the project manager who runs the board and no one else is allowed to touch it. And then they just tell everyone what they're working on and when everything is due. Which is how that, you know, your AI surrogate that goes between and then eventually replaces also would work. I just, where I'm, you know, we've been saying this for, God, how many years now, Josh? Been a few. 10 years ago, I think you could go back and there's a version of this conversation where we're like, really soon. Yeah, yeah, yeah. Oliver, because you look at Intuit and you're like...

Why wouldn't most software just be conversational? It's on its way, and then it doesn't come this way. And I think there's just so much money in antiquated software, just like there's so much money in oil. Yeah. It's also familiarity, like it is, the thing you're saying, resistance. Everyone hates Slack and loves Slack. The version of Slack that you can map onto what you're saying is...

you know tell john i've done the thing and it just does yeah and um but what does it take do you think to change your neighbor's mind like what is well i don't know i think people do like typing at each other yeah in that example like i do i think people do like you know the like that's a little gamify there's there's emojis there's different threads like like it it that that software also like is a surrogate for sitting in an office sometimes you know i think that social people like so um and then yeah maybe my neighbor like maybe, maybe he likes that part of his job. So like, you know, some people do like the, the part of a job that other people don't, you know, like, um, I, like I was a civil engineer a million years ago. That's why I studied in college and I had a very, um, I had a job where I counted traffic. It was like an entry-level summer. I sat at intersections and punched buttons as cars drive by and then loaded it into the computer and printed out the reports. And then we had to go down to the thing and file the intersection of Edgecombe and Grant Street report, whatever, peak traffic volume or whatever. And I hated it, but I worked with a guy who loved that part. He's like, I love just laying them all out, alphabetizing them, sorting them, putting them in the thing. It's a thing.

Yeah, it's funny. It's funny you say that. That's the whole AI replacing doves. Do we need that guy to do that? Absolutely not. There's a much easier way, a much cheaper way to store this stuff. People have argued that email is a functionally dead platform in many ways, and it's so...

choked by spam and choked by, you know, like, and choked by automated stuff and choked by onboarding and all these different things. And yet, like, you know, I'm not an inbox zero guy. I just still have like things from four years ago. That's like a nice letter from a teacher I had in grad school. And it's like, like the parts of it that are familiar and tactile and feel like a letter or, you know, the part about Slack that feels like a water cooler conversation. Like people want, sometimes I think people want UI, like they don't want it to completely go away because the parts of UI that do mimic familiar and comfortable parts of life.

Ed Zitron, the PR guy who has a really good blog about workplace stuff. He writes a lot about return to office and how the reluctance of that is like a middle management class that kind of actually like... like we haven't reached peak efficiency as a society or as a working society but like you know if it's true that bank of america has 10 000 too many employees at any given time maybe if that's true there's a lot of people for whom like like my example like people like slacking each other people like receiving emails from friends or hold on to them because there's like a connection to attack the workplace like his proposal was that like return office was just because like if boss can't see you doing stuff or a middle manager can't like middle manage you know like they don't have anything to do yeah like there's like my work as a copywriter is like continuously visible because like even when this interview is over i like have to i have to send someone like an update to a web page and like someone sees that but the person who approves it like they don't their work product is just like the correct go through and and so like i wonder if um there will be like an also intractable reluctance to an option of of tools that make the work invisible because there is like

whether whether you're a person who like finds great dignity in your work or extracts lots of power like positive negative whatever the reason like people like going to work like yeah you know you're both remote employees and she like really misses living in new york and like getting on the g switching to the l and going to midtown like for her job and she has the same job she used to have and like i wonder if like like the i think there's like a couple things to unpack there the first one is i used to call this i still do the scarecrow effect um and i used to say a good manager has to like go above and beyond the scarecrow effect and the scarecrow effect is like is more physical like you're in the office you're the manager and then orbiting you are people and they know you're there you're just like doing you're playing solitaire on your computer right but they know you're there so they're all working because at any moment you could come over the scarecrow like oh are you working like so so they sit there cranking away like all of a sudden the manager gets up and is sick that day and nobody like productivity goes down because so what is the value of the manager like just the scarecrow like it doesn't matter if they do anything just their just their sheer presence

has an effect. And I was like, I don't want to give anybody raises. I don't want to hire people for just simply being a scarecrow. And you're like, well, what happens when it's remote? How do scarecrows scarecrows? Well, it seems like Microsoft Office is becoming the scarecrow, right? It's with Slack, right? It's with Slack and Teams. Yeah, see who's online. Are you online? Watching your little status and then seeing all the conversations. So now you've got the virtual scarecrow effect.

where you got these managers popping in to different conversations. Like, I'm watching. But now you got to like actually be talking to show you're doing work. Like if you're not communicating, like you as a writer, if you're just heads down for half a day, how do they know that you're heads down half a day? Unless you're like talking to someone about it. Yeah, I definitely know Instagram might walk in the woods when I'm online for half a day. Right, yeah. So what are we doing at this point? Like productivity is...

This is sort of the unpacking of what you said, but at some point you're like, well, why do you go to work to be productive? Well, what if productivity isn't the measure, it's creativity? How good was the output? It doesn't matter how long it took you to get there. It doesn't matter how much time you put in. It matters how good it is. And if an AI could actually just look at your end product and grade it and say like, from a creativity standpoint, how good is this? Or a person could, or a combination of AI and people, and we just get graded on the output, then ultimately...

work becomes about what? About interacting with other people and machines. This kind of all comes back to like, if you fall in love with a machine like her, right? Everybody talks about that. Is that a life well-lived or does it have to be a human? If your coworkers are AIs and you're having a conversation with AI or you're having a conversation with other people, does anyone care? Does your boss care?

Right. Yeah. It's, um, I think your boss is like, cool. I only had to pay one. Like when you talk to chat GPT to come up with it, like I only had to pay one person. Yeah. Like I got an MFA and when you would put up stories at your MFA, you know, they put the current of silence on you and you kind of have to sit there and listen to like your peers talk. Like it's, it's worse than death. You know, it's a single most excruciating hour of your entire year. And you repeat it a couple of times every semester.

and i just had one professor who was like i think everybody oh and one of the rules was like you don't really like you critique the work like sort of new criticism you just like treat it for what's here you don't ask questions you don't make suggestions you just like talk about it and the person has to watch the room talk about it like you know right huck been watching his or tom and huck watching their own funeral um and like one of my professors was just like i am allowing everyone to make suggestions because i read a thing about old hollywood how they'd always employ like one crank they'd have like seven writers in the room and then one like they call him crazy guy and he'd be like what if we you know do this and he would stretch the the idea past the point of feasibility or acceptability or decorum and then like that stretching would then give everybody else in the room the room to walk further up to the line like that is like and you know permission to be creative essentially yeah and that was like that's the um that's where like i can do that with my buddy when we send each other work and i haven't found the model that can that can give you the like

right hallucination, like, you know, use the language model word, like, at the right time. As adept as the models, from my perspective, aren't, like, solving, problem-solving copywriting. When I'm asking it to do something a little more intangible for me, I get, like, not the version of that yet. Yeah. I think it's just, if you think of what an LLM is, it's compressed knowledge, right? Which means normalized. So, can it...

do outliers absolutely there's like there's like settings you can get the the unlikely the problem is just like evolution the more mutations that you add most of those mutations are going to be like bad oh they're going to be they're going to kill the like most mutations end up in death of the organism right at it then end up in an evolution right so this is the problem with with llms evolving is they have to be able to make a lot of mistakes And, and, and that might be at our expense. So when you think of like customer service, right? You're like, Hey, you overbilled me and they come back and LLN comes back and says, well, great. I'm going to overbill you again. Like, it's just going to try different combinations to find that outlier mutation that actually works. So I think like that, yeah, that.

What it has to do, and that's just, I think, in the design, what we don't have is the close loop. What you can do is come up with a crazy idea. And you said it, the guy comes up with a crazy idea. The rest of the room, it's a simulation, right? Everyone's running a simulation now. They run the joke, or let's say it's a humor, because I think that happens a lot with humor writing, right? You got the guy who goes way out there, the Mel Brooks in the room, that comes up with something. And maybe it's too out there to be funny. And everyone's like, no, that's not funny. What they're doing is they're running simulations. If the AI can run simulations, then it can work. Those outliers can happen.

But it has to be like, okay, now I want you to run up a bunch of crazy ideas in rapid succession and then evaluate the result and come back with the ones that are actually funny. And we haven't seen that yet because we haven't prompt chain to that degree. We know it can evaluate. We know it can do a decent job of evaluating if something's funny or not. But you just haven't gotten to that place where you're like, hey.

come up with crazy ideas and and only the ones that actually might work yeah yeah exactly i worked at bark box the dog toy treat like the dog um d2c company and we like part of my job was naming dog toys like me and another company and we just like it was all puns it was all dad jokes it was like That was not just what the job was, but that was like the very best part of it. And like our boss just, he just would be like, all right, here's the list. Like here's whatever, here's the voyage to Italy box or whatever. And we just sit down after, you know, a week later, once we decided the theme or looked at the toys and just, just throw names at it. We would just do that. It was just, every single one was an audition until one made him laugh. And it was like, I don't know how to teach understanding my boss, Nick's vibe to anyone, but I know what will make him laugh and what will I get the, you know, but that's, well, that's one of those things too. It's like, do we want to, oh, go ahead. We had this earlier episode where we came up.

kind of like future of work idea. And it's starting to coalesce like one big part of future of work in terms of productivity, like not in terms of socializing. Is this like hot or not like decision? We're just going to be the human in the loop decider. Yeah. And so you're just going to get paid to say, is this funny? Is this funny? Is this funny? Like swipe laughs, swipe white. And so AIs are going to come up with crazy ideas and they're going to pay you to evaluate them. And then, because they themselves may not be.

a good enough judge of whether it is or isn't. Or maybe they're just getting trained to be able to kind of predict what's funny or not. And so you'll just sit there like, our day will just be swiping left and right on this. Do you want, what kind of, I'm going to buy this ink for your printer. Good to go. Yep. I'm going to, hey, is this funny? Yes, no. It's like, we're all like mechanical turk.

getting like micropayments for AIs just running stuff past us. That's not, not a creative director's job. Creative directors also do other things, but that is like, that is what my CD does. Right, right. Like with another one, like come back with, eh, not there yet, you know? Yeah. And so, so if you're just, if you're just the layer before him, like, hey, can you cut through the first layer and then narrow it down to 10 choices? And then, and then he gets.

he only gets 10. You had a, you had a hundred that you weeded through. So you got 10 from you. He is in jeopardy. Or, or your job just becomes like, yes, it tastes good taste. Like you're just like at the end of the day, that's what you are anyway is, is, is running scenarios in your own head and then saying, this is the one I, yeah, this is good. So maybe, maybe your job is good taste, not the creating of it. And, and so you're just at the end, the idea of like all of the craftsmanship around.

getting to a product you can evaluate as good or bad. Like that shortens. And now you can, it's like a painter that like paints over the same painting a hundred times. Like the canvas is just thick with paint because they just kept trying different things until it was good. And then the bad painter only tried three things and then gave up. Well, yeah, to push back though, how do you get that? If the models replace, you know, if we're just swipe left and righters, how do you gain, like I would argue that you gain some of that taste through the process of doing that work that you would train your model to do for you.

Yeah. It's like a chicken or the egg. Like, what if the work that you're sitting there, that's routine, is part of the creative process itself. Right. Yeah. To bring it back, I'm going back to this flagstone walkway I'm going to be doing over the next week at my house. Right. I've gone on Reddit and people were like, hey, here's my flagstone walkway. What do you guys think? Because it's someone who's done it for the first time. And a common response is, well, what do you think? It's your house. Yeah. I think that's not expertise. They don't know if they did a good job or not. Yeah. Like, what if the laying of the brick in that slow, methodical...

boring process is your creative process. I mean, I'm very excited. And you need to lay brick. Yeah. But maybe that's how you come up with the next writing thing. Like by laying brick is how you think. Then I think there's no danger of AI. In a way, we've like circled the conversation. Then you're going to lay brick to come up with the next good idea. And you're not going to ask AI to lay brick for you because then ultimately you're not going to do your job. And so we'll find that...

The parts of your job that are routine that don't lead to your creativity and good taste, you'll hand to an AI. And parts of your job that are routine that you do need that feed your creativity, you will do. But at the end of the day, you'll choose. You'll be the decider of what you want AI to do and what you don't want it to do, not your boss. Because ultimately...

Your boss is just going to want the creative output. And they're going to stop caring how you get there. We definitely appreciate you taking the time with us today, Matt. It's been really fascinating. That's great. We barely talked about Workday. That's true. Yes. So I think the idea is these old UIs and how quickly they're going to disappear because AI will operate them.

Yeah, I think that's what fascinated me when I read the article. A couple of things. First, you talked about Workday, but you also did a similar piece on Asana. And then I started beginning to see like, well, both of these will be the same UI someday. And are they going to be like two crappy UIs that we Pogo stick between like they are today? Or are we going to see a whole new UI? And will it be Workday and Asana that create it or will it be a new company?

That disrupts both of them. On record, I'm fine with anyone replacing any of them with anything else. I am old man yelled at loud when it comes to all of them. Yes. Excellent. Well, thanks again, Matt. I appreciate it. Yeah, this was great.

All right. Well, thanks for hanging out with us again here on Invisible Machines. Don't forget to like and subscribe wherever you get your podcasts. Always remember that you can watch this podcast, too, on our Invisible Machines YouTube channel. Head over to uxmag.com. We have a podcast page there where you can explore our back catalog and you can look at episodes centered around popular topics. Many, many thanks to our incredible production team here in the U.S. and in Ukraine, Elias Parker, Michael Lipinov, Allison Harshberger, Alis Lesarenko, Vera Pakhodkov, and Dorena Moskovchuk. Till next week. Owaki.

Frequently asked questions

How long is this episode of Invisible Machines podcast by UX Magazine?

This episode is 1 hour and 4 minutes long.

When was this Invisible Machines podcast by UX Magazine episode published?

This episode was published on September 12, 2024.

Is there a transcript available for this episode?

Yes, a full transcript is available for this episode. You can read the complete transcript on the episode page.

Can I download this Invisible Machines podcast by UX Magazine episode?

Yes. Use the download control on the episode player to save the publisher-provided media file.