Latest / Logixly Live Unsupervised Learning Podcast / Exploring the AI Adoption Funnel

Transcript

Exploring the AI Adoption Funnel · Logixly Live Unsupervised Learning Podcast

• 0:00
  SPEAKER_01: Welcome everybody where we try to cover uh whatever's happening in AI.
• 0:03
  SPEAKER_01: And of course it's always changing week to week and there's always something new to talk about.
• 0:08
  SPEAKER_01: And speaking of something new uh to talk about, we got somebody new to talk to, which is Ashton.
• 0:14
  SPEAKER_01: Ashton, you want to do a quick intro of yourself?
• 0:17
  SPEAKER_02: Yeah, I'm uh one of the experience designers here at NLP Logixly, and I work on a lot of what the people see, what they experience, what they look at, what looks back at them sometimes.
• 0:28
  SPEAKER_02: A little bit of that.
• 0:30
  SPEAKER_01: Which uh I know this isn't quite to script, but I gotta ask.
• 0:33
  SPEAKER_01: Experience design, is that that uh has that just totally changed then in the post chat GPT era?
• 0:40
  SPEAKER_01: Is it or is it totally changed?
• 0:42
  SPEAKER_02: Uh you know, it it'd be I think it would be advantageous to to say yes.
• 0:47
  SPEAKER_02: It's it certainly has.
• 0:49
  SPEAKER_02: It's all different now, but really it's kind of not.
• 0:51
  SPEAKER_02: I mean, it kind of it's made things easier, like probably like a lot of professions, but at the same time, it's it's it still comes down to some fundamentals, you know, okay that drives the experience mostly around humans.
• 1:03
  SPEAKER_02: Like we're kind of the same.
• 1:04
  SPEAKER_02: AI has made us work a little differently, so we gotta adapt to that.
• 1:07
  SPEAKER_02: But deep down we're kind of the same.
• 1:08
  SPEAKER_02: So is that experience that we have to design for?
• 1:11
  SPEAKER_00: All right.
• 1:12
  SPEAKER_00: See Ashton sitting in my tune, because that's what that's my play.
• 1:14
  SPEAKER_00: It's always the human, it always comes down to business and humans, and how do we use tools?
• 1:20
  SPEAKER_00: That's right.
• 1:20
  SPEAKER_00: And uh and Ashton is absolutely amazing, uh, quite the talent.
• 1:25
  SPEAKER_00: We've had a lot of clients.
• 1:27
  SPEAKER_00: Very nice.
• 1:27
  SPEAKER_00: Yeah.
• 1:28
  SPEAKER_00: No, we've had a lot of a lot of client comments talking about how Ashton really broke things down from a um UX standpoint, and the rest of the UI UX team, of course.
• 1:39
  SPEAKER_00: But uh yeah, so it's great to have them here.
• 1:41
  SPEAKER_01: It is, and and I think people who know Ashton, I uh the AI collaborate our annual conference last year.
• 1:47
  SPEAKER_01: I just feel last year's was uh I think we all I remember it, I think it was fantastic, but just just the way to communicate really elegantly, uh interesting topics, it was a lot of fun.
• 1:58
  SPEAKER_02: Yeah, and if you want to experience it yourself, I noticed our new page for this year is up on our website, and there's a picture from last year, which I did not approve of, by the way.
• 2:08
  SPEAKER_02: But it is it will sum up the experience if you just look at that picture.
• 2:14
  SPEAKER_02: So we'll see.
• 2:15
  SPEAKER_01: Well, today's topic.
• 2:16
  SPEAKER_01: So speaking of like uh change, I guess, um couple of couple of topics I think would be fun to hit today.
• 2:23
  SPEAKER_01: One is the the first one is uh Carpathy's uh post that he had on on X a few weeks ago, or maybe about a week and a half ago, but basically saying, you know, if you if you don't know Carpathi, he's uh you know, I'll paraphrase some of his background.
• 2:40
  SPEAKER_01: Like he's certainly one of the luminaries in in AI.
• 2:43
  SPEAKER_01: He was Tesla's, this is off the top of my head, but he was Tesla's head of AI for certainly a long time, made a lot of the progress on full self-driving.
• 2:53
  SPEAKER_01: Then I believe he went to the one of the frontier labs.
• 2:56
  SPEAKER_01: He was at OpenAI for a bit.
• 2:57
  SPEAKER_01: Um, and anyway, obviously when he says something, I think people listen.
• 3:02
  SPEAKER_01: You know, uh he made some good good feedback or good points on vibe coding.
• 3:07
  SPEAKER_01: He um had this auto research tool that was kind of showing how AI could kind of self-improve AI.
• 3:13
  SPEAKER_01: And then now, you know, he's got this this talking point around the funnel around AI adoption, which I think is really interesting because it is it's one of those things to me, it's it's you experience it maybe over the course of weeks and months within interactions that we have with other um other groups or customers' partners.
• 3:36
  SPEAKER_01: But then you see, you see someone else explain it, and you're like, that's the concept.
• 3:41
  SPEAKER_01: So Carpathy's funnel is basically the the two edges, right?
• 3:47
  SPEAKER_01: Where it's like you have look, most of the world, but even even the US, and you know, just like most people do not interact with AI every day.
• 3:57
  SPEAKER_01: Uh, and then there are people that do, but they interact with AI in kind of superficial ways, refine this email, help me with a recipe or something.
• 4:08
  SPEAKER_01: Like in these ways where it's like, yeah, that's AI.
• 4:11
  SPEAKER_01: It is, it is just tip of the iceberg, right?
• 4:15
  SPEAKER_01: And then on the other extreme, you have, and Carpathy puts it at like 5,000 people, but give or take, basically at these labs that are working with versions of AI that are actually probably not released, like the frontier, the front edge of the frontier.
• 4:31
  SPEAKER_01: It's it's the most cutting edge.
• 4:34
  SPEAKER_01: And how uh when you talk to somebody, you really don't, you really might want to know where they're at on that uh spectrum as it relates to whatever the conversation is.
• 4:45
  SPEAKER_01: And so I think you know, one of the interesting things I think is like when you're talking with somebody and somebody has that opinion of is AI hyped, overhyped, underhyped?
• 4:56
  SPEAKER_01: It's like you gotta keep in mind that there is a massive spectrum of where are we in this in this world?
• 5:05
  SPEAKER_01: I don't know.
• 5:06
  SPEAKER_01: Uh any thoughts on on anybody surprised by Carpathi's thoughts or or does it put anything in a different perspective than what you guys have been seeing?
• 5:16
  SPEAKER_00: No, I think he's spot on.
• 5:18
  SPEAKER_00: That's why he has a conference room named after him here at uh at NLP Logixly.
• 5:24
  SPEAKER_00: But you know, I've I'm always you know, my angle is always because I'm not technical, is always the business side and the adoption side.
• 5:32
  SPEAKER_00: Like you can build the greatest applications or things in the world, and if no one uses them, who cares?
• 5:38
  SPEAKER_00: And and so I really do think it's it's the divide isn't technical because whether you're a not a super user, just using it to rewrite an email, or whether you're you're you know, on the on the cutting edge, everyone's using the same tech.
• 5:55
  SPEAKER_00: It's just how are you using it differently?
• 5:57
  SPEAKER_00: And and um, you know, is it part of your organization, is it part of your operating model, or is it just kind of something that's off to the side and and there's just that that huge chasm.
• 6:07
  SPEAKER_00: I mean, what do you think, Ashen?
• 6:09
  SPEAKER_00: You're you're yeah, you're probably a super user, I think.
• 6:12
  SPEAKER_02: You're up the you're not on the frontier because you're here, not on the frontier, but you know, I dabble, I've been known to dabble.
• 6:18
  SPEAKER_02: But I think it comes down to two things that's crazy because it it we don't know how people are gonna use it.
• 6:25
  SPEAKER_02: So, you know, I think a lot about some employees, you probably you may have those employees who have like this crazy keyboard setup, you know, they reject the default thing, they bring in their own keyboard, their own mouse, and it does these crazy things and these macros and stuff, foot pedals, they've got a steering wheel, some of them they're like, Oh no, I'm just you know, browsing Google, I'm not driving.
• 6:43
  SPEAKER_02: But imagine somebody came to your desk and dropped off this equipment and said, This is gonna help you be very, very efficient.
• 6:49
  SPEAKER_02: Trust me, it's gonna work great.
• 6:50
  SPEAKER_02: Everyone's doing it, it's the latest thing.
• 6:53
  SPEAKER_02: Okay, see you later.
• 6:54
  SPEAKER_02: I gotta go.
• 6:55
  SPEAKER_02: So, I mean, you're gonna struggle figuring out how does this these tools I know they can work, they're working for other people.
• 7:01
  SPEAKER_02: How can they work for me?
• 7:02
  SPEAKER_02: And that's kind of the struggle, is where it comes down to people, experience design, sit with them, what are they doing?
• 7:07
  SPEAKER_02: I wanna be micro, dirty jobs.
• 7:09
  SPEAKER_02: I want to know what you're doing, what you're struggling with.
• 7:11
  SPEAKER_02: I want to hear your hesitations.
• 7:12
  SPEAKER_02: If I'm training you to use AI, I want to hear, oh, okay, maybe I should reword this.
• 7:16
  SPEAKER_02: Or that wasn't quite the output.
• 7:18
  SPEAKER_02: These little things that would have made somebody stop normally, if you're not there to listen to them, you can be there, you can participate, you can hear, and you can say, Oh, this is what you're doing wrong.
• 7:26
  SPEAKER_02: And then once they have that aha moment and they realize, wow, this actually is really helpful.
• 7:31
  SPEAKER_02: Turns out I can browse Google with the steering wheel.
• 7:33
  SPEAKER_02: This is super fun, actually.
• 7:34
  SPEAKER_02: You know, they're gonna be addicted to it, and maybe it'll help them out in more ways than they could have imagined.
• 7:39
  SPEAKER_01: And and I I feel like the sentence that I love around that is the license is not a capability.
• 7:46
  SPEAKER_01: As in, I mean, you you put that keyboard down in front of you, you have that license now, but it doesn't mean you're more capable at what you're doing.
• 7:55
  SPEAKER_01: Right.
• 7:55
  SPEAKER_01: I mean, is that that's that's that's that's getting to the root of it, right?
• 7:58
  SPEAKER_02: Right.
• 7:59
  SPEAKER_02: And it it's it's hard with those adoption metrics too, because what do you look at usage, like usage, licenses, outputs, yeah, and stuff like that.
• 8:06
  SPEAKER_02: And it's interesting because there's actually a rose, it's called the thousand year rose, it's in Germany.
• 8:11
  SPEAKER_02: It's on this cathedral.
• 8:12
  SPEAKER_02: Have you ever heard of this?
• 8:13
  SPEAKER_02: I have not.
• 8:13
  SPEAKER_02: Supposedly it's been alive for a thousand years.
• 8:16
  SPEAKER_02: Thousand years.
• 8:17
  SPEAKER_02: Can you imagine that?
• 8:18
  SPEAKER_02: Wendy's shocked back here.
• 8:20
  SPEAKER_02: A thousand years.
• 8:21
  SPEAKER_02: I'm surprised.
• 8:21
  SPEAKER_02: I'm back.
• 8:22
  SPEAKER_02: We have roses on the table.
• 8:24
  SPEAKER_02: I'm shocked, wondering where you're going.
• 8:27
  SPEAKER_02: In 1945, he's going somewhere in World War II.
• 8:31
  SPEAKER_02: It was decimated.
• 8:32
  SPEAKER_02: You know, the bombs are dropping.
• 8:33
  SPEAKER_02: There's a lot going on, obviously, and it was wiped out.
• 8:36
  SPEAKER_02: But it grew back.
• 8:37
  SPEAKER_02: The rose is still there to this day, better than ever.
• 8:40
  SPEAKER_02: But there's a debate.
• 8:41
  SPEAKER_02: Is it really still the thousand-year rose?
• 8:43
  SPEAKER_02: Because it was the root system that it that maintained and existed, and it's now grown back and created a new rose.
• 8:48
  SPEAKER_02: So it's funny because we look at all these adoption metrics, and I think a lot of companies have these rose pedals.
• 8:54
  SPEAKER_02: You can see all of these the outputs we talked about, the token usage, but they're not looking at the the root system, okay, which is the actual how are what we just talked about.
• 9:03
  SPEAKER_02: How are you actually using it in your workflows?
• 9:05
  SPEAKER_02: Are you using some of the agentic flows that are in there?
• 9:07
  SPEAKER_02: What's that?
• 9:08
  SPEAKER_02: Never heard of that.
• 9:09
  SPEAKER_02: You know, are you using custom GPTs and now those are gone?
• 9:11
  SPEAKER_02: You have plugins, your dot, using your dot, you know, these those that's the root system.
• 9:16
  SPEAKER_02: That's the thing that's gonna survive and really thrive.
• 9:18
  SPEAKER_02: Everyone can have a whole bunch of pedals and right eventually.
• 9:22
  SPEAKER_01: You gotta go.
• 9:23
  SPEAKER_02: Right, you've got to like that.
• 9:24
  SPEAKER_02: That's where the that's where the real stuff is.
• 9:26
  SPEAKER_02: Is that the root system?
• 9:27
  SPEAKER_02: That was deep.
• 9:27
  SPEAKER_02: That was that was good.
• 9:28
  SPEAKER_02: How far?
• 9:29
  SPEAKER_02: I mean, that was a deep, it's a root system.
• 9:32
  SPEAKER_01: So that's good stuff.
• 9:33
  SPEAKER_01: All right, I gotta ask you, because you brought up the um the keyboard question too.
• 9:38
  SPEAKER_01: So kind of a related question.
• 9:40
  SPEAKER_01: I I find it real I'm really interested in the thought, and especially since your your background of you know, experience design, really interested in the thinking of is today's offices, computers, you know, we have keyboards, mice, we have, we click buttons, we swipe.
• 10:01
  SPEAKER_01: And all of those interaction paradigms were really before we have like intelligence as a service.
• 10:08
  SPEAKER_01: And now we have this omnibox chatbot, which to me, I don't know, it seems very coarse-grained.
• 10:16
  SPEAKER_01: And uh just my opinion, but like the you know, if it's chat GPT Enterprise or whatever, like these tabs and it's chat or it's co-work and the like it just doesn't seem yeah, it seems like there's a better way.
• 10:31
  SPEAKER_01: I'm very curious, Ashton.
• 10:32
  SPEAKER_01: What like a voice to me?
• 10:33
  SPEAKER_01: Like, I feel like I I don't type well.
• 10:36
  SPEAKER_01: Ted knows this.
• 10:37
  SPEAKER_01: I don't spell well, I don't type well.
• 10:39
  SPEAKER_01: A lot of things I don't do well.
• 10:41
  SPEAKER_01: Uh, but I like to talk and I I like to, and I feel like that that's a more natural way for for an interface to something.
• 10:51
  SPEAKER_01: I'm just really curious.
• 10:52
  SPEAKER_01: What do you think about the ways we interact with AI today?
• 10:55
  SPEAKER_01: Where does that does it feel right to you?
• 10:58
  SPEAKER_02: Uh that's so hard because it changes so much.
• 11:01
  SPEAKER_02: Um, but I mean, I don't even know that talking feels right.
• 11:04
  SPEAKER_01: Okay.
• 11:04
  SPEAKER_01: You just want to think it.
• 11:06
  SPEAKER_02: I just want it to be telepathic.
• 11:08
  SPEAKER_02: But really, though, right?
• 11:09
  SPEAKER_02: Because it knows what's on my computer, it knows my emails, it knows what I need to prep for, it knows stuff before I know it, because I haven't had a chance to look at this document that's super long.
• 11:18
  SPEAKER_02: So I kind of wanted to start anticipating what I need and getting a jump start on it.
• 11:23
  SPEAKER_02: So it's even better if I don't have to say anything to it.
• 11:25
  SPEAKER_02: But of course, you know, there's a lot of course correction that would come with that.
• 11:28
  SPEAKER_01: Well, I you know, but I mean, to your point, and I think I shared this last week, like uh these last couple weeks I've had two like this is what I wanted AI to be for me moments.
• 11:37
  SPEAKER_01: One of them was my dot.
• 11:39
  SPEAKER_01: So I I got an email, I hadn't read it yet.
• 11:42
  SPEAKER_01: And really before I read and processed the email, the dot threw up a highlight that said, Hey, that Excel sheet you just got in your inbox, tab two.
• 11:51
  SPEAKER_01: I think there's a formula error on it.
• 11:53
  SPEAKER_01: Here's the details.
• 11:55
  SPEAKER_01: It anticipated like to me, yeah, I was like, this is that's kind of an experience I want to have.
• 12:00
  SPEAKER_01: Like I that's nice, yeah.
• 12:01
  SPEAKER_01: That was right.
• 12:02
  SPEAKER_02: Felt like you someone is in your corner a little bit.
• 12:04
  SPEAKER_01: It felt like it was anticipating and it felt like it knew what I was gonna do.
• 12:09
  SPEAKER_01: And it gave, you know, it was helpful.
• 12:11
  SPEAKER_01: It was, hey, by the way, just a little check.
• 12:14
  SPEAKER_02: Right.
• 12:14
  SPEAKER_02: And those dots are the new the AI assistance kind of that open AI is rolling out.
• 12:18
  SPEAKER_02: If you have chat GPT, the later plans you you'd see some of those.
• 12:21
  SPEAKER_02: But I think that solves some of those problems because right now part of my problem is I have to open up a new chat.
• 12:26
  SPEAKER_02: I have to decide work mode or chat mode, then I have to decide which plugin to use, and then I have to I don't want to decide which model.
• 12:33
  SPEAKER_01: Like that to me, the models the models are email is darn important.
• 12:36
  SPEAKER_01: Like I don't want to say give me the worst I get it from a cost standpoint, yeah.
• 12:41
  SPEAKER_01: But what is it?
• 12:42
  SPEAKER_01: When do you want to say I want to deliver a mediocre email?
• 12:46
  SPEAKER_01: Even if it's email, like I you want to take ERCOT down.
• 12:50
  SPEAKER_00: Writing that email with the data centers burning up and texting.
• 12:54
  SPEAKER_00: It's an important email.
• 12:55
  SPEAKER_00: It's take belt.
• 12:56
  SPEAKER_02: They tried to address that by kind of removing that from the UI and then saying, oh, we're also gonna switch to the best model based on kind of what you're doing.
• 13:03
  SPEAKER_02: So they're they're trying to address that, but you're right, there's still all these little, there's too many decisions to be made, which is also why I like the dot though, because I just tell it what I want it to do, and it's gonna go, it's gonna open the chat, it's gonna run whatever plugins it needs, it's gonna let me know if it needs a login or a password somewhere, and I'm just gonna type it in without even acknowledging.
• 13:19
  SPEAKER_02: Um no, of course, you're gonna check your passwords where you're logging in.
• 13:22
  SPEAKER_02: But it's gonna do all this stuff for you, and you don't have to like go through that the mental bounds of the chats, and uh stuff's getting lost in my chats too.
• 13:30
  SPEAKER_02: And I'll tell go find, hey, go find this file from this chat last month.
• 13:33
  SPEAKER_02: So there it goes.
• 13:34
  SPEAKER_01: Okay, so here I wanna so Anthropic put out a really okay.
• 13:38
  SPEAKER_01: I love the research that comes out.
• 13:40
  SPEAKER_01: Uh OpenAI does this, Anthropic does this, and and we have a visual on this that we can show, and it's basically talking about all of the output.
• 13:49
  SPEAKER_01: If you think of all of the tokens that are being created and turned into words for us and stuff that we can consume, it was um it was this graphic, and I'll hit the top left that says um in the engineering and research function, so tokens that are used by individuals from the engineering and research role, on average, uh it's generating 400 hours of work a week for um for each individual to consume.
• 14:21
  SPEAKER_01: Okay.
• 14:21
  SPEAKER_01: And that in the in the in the chart on this, this is basically a uh that's actually only in April.
• 14:27
  SPEAKER_01: So that's from December to April.
• 14:29
  SPEAKER_01: December, it was maybe around 50 hours, and then it 10x over the course of the year.
• 14:34
  SPEAKER_01: But to and I'm curious, you guys, how many, how many, and I'm bad at this, how many chats are you like, uh I'm gonna fire off this work.
• 14:45
  SPEAKER_01: I wanna do this, go make some progress on it, and then you don't come back, or you forgot to come back and to consume the output of it.
• 14:55
  SPEAKER_01: Like, are you guys do you guys ever have that experience when you're like, wait a second, I I wanted to make progress on this.
• 15:02
  SPEAKER_01: I asked, I delegated to my agent, and then whatever the output was, it's either lost because I've had 10 other threads since then, or it's removed, you know, it went out of my context window basically.
• 15:14
  SPEAKER_00: So I had that two weeks ago.
• 15:17
  SPEAKER_01: Okay.
• 15:17
  SPEAKER_00: Um writing a proposal.
• 15:20
  SPEAKER_00: So, you know, I love AI from that standpoint.
• 15:23
  SPEAKER_00: It can it can see all past proposals and where we've interacted with the client, and it's it's awesome, but it's chewing through a lot more information.
• 15:33
  SPEAKER_00: You know, put in the prompt.
• 15:34
  SPEAKER_00: My prompts are not very sophisticated, but this was gonna be one that was gonna chew up some time.
• 15:39
  SPEAKER_00: You know, wrote the prompt, hit hit return, and then immediately locked my screen in case the security guys are listening, and then walked off, and like two and a half hours, two and a half hours later, got back to my desk, get a phone call.
• 15:55
  SPEAKER_00: Hey, can you come downtown, meet on something?
• 15:57
  SPEAKER_00: Okay, I can do that.
• 15:58
  SPEAKER_00: Yeah.
• 15:59
  SPEAKER_00: And like two days later, I come back, I'm like, oh yeah.
• 16:01
  SPEAKER_00: Oh right.
• 16:03
  SPEAKER_00: Yes.
• 16:04
  SPEAKER_00: Uh, and I luckily I didn't miss a deadline.
• 16:06
  SPEAKER_00: But it was, but it was one of the exact things we were talking about where yeah, I just I burned a bunch of stuff and did nothing with it.
• 16:12
  SPEAKER_01: Yeah, and and I I I feel like the the to me, the subtext on what Anthropic was saying was if we are well, two interesting one points.
• 16:22
  SPEAKER_01: One was for their engineers, their data point was uh even though we're generating so much output per per resource, per person via agents, agents are mostly idle for for a work day.
• 16:37
  SPEAKER_01: So even though it's creating all this stuff, they're mostly not doing things.
• 16:40
  SPEAKER_01: That's the worst.
• 16:41
  SPEAKER_01: They're waiting on you.
• 16:42
  SPEAKER_01: They're waiting.
• 16:42
  SPEAKER_01: So, and that was the point of it.
• 16:44
  SPEAKER_01: So it's it's not only maybe they're waiting on you because it's generating so much output that you have to go and adjudicate, right?
• 16:52
  SPEAKER_01: And then the second part was what is the what is the layer on top of this output that we need to have to help orchestrate this work?
• 17:03
  SPEAKER_01: So is it is it is it going are you know, right now we're basically going right from output to human.
• 17:10
  SPEAKER_01: Is there another layer, which to me is I hear this and I'm like, man, this is so meta and abstract, and it just it but it's like it also is real.
• 17:19
  SPEAKER_01: It's like you got 400 hours of output tokens to deal with this week.
• 17:23
  SPEAKER_01: How do you manage that?
• 17:25
  SPEAKER_01: How do you prioritize?
• 17:26
  SPEAKER_02: What is the you either get help or you reduce the output?
• 17:29
  SPEAKER_02: Right.
• 17:30
  SPEAKER_02: So either have an AI help.
• 17:32
  SPEAKER_02: What would you talk about?
• 17:32
  SPEAKER_02: The meta, have AI, watch AI to watch AI, or you get better at those outputs, you know.
• 17:37
  SPEAKER_02: Don't say, create me a 50-page report.
• 17:39
  SPEAKER_02: Say, you know, find three contradictions that were in this user feedback session and the top three reoccurring uh issues that users are experiencing.
• 17:48
  SPEAKER_02: Write down any questions I should ask my PMs, you know, get better at the prompts.
• 17:53
  SPEAKER_02: Ted.
• 17:53
  SPEAKER_02: PM?
• 17:54
  SPEAKER_00: This prompt?
• 17:55
  SPEAKER_00: Get better at the at those prompts.
• 17:59
  SPEAKER_02: No, I'm just saying either.
• 18:00
  SPEAKER_02: I mean you have to problem to solve.
• 18:02
  SPEAKER_01: I mean, I and so another experience, again, maybe this is just a madism, but the other one is um is I'll get uh a good output that I like like 40% of, and then but it's it's it's sick, which means it's 60% of like junk slash garbage.
• 18:19
  SPEAKER_01: And and I I also feel like iterating on the good 40% and actually getting it narrowed and down to so you know, I don't want the 10-page version.
• 18:29
  SPEAKER_01: I want to I feel like I'm always constantly trying to simplify and winnow down.
• 18:34
  SPEAKER_01: I don't know what you guys have for experiences in that way, but um it's just I'll go back to my point of of it feels Ashton, I'll put this on you.
• 18:43
  SPEAKER_01: It feels like the experience could be improved.
• 18:45
  SPEAKER_01: I don't know what you're gonna do about it.
• 18:47
  SPEAKER_02: You're right, yeah.
• 18:48
  SPEAKER_02: I've got I've got a lot of work to do.
• 18:49
  SPEAKER_02: That's the one thing I'm taking from this conversation.
• 18:52
  SPEAKER_02: A lot of eyes on me.
• 18:53
  SPEAKER_02: But you're right, yeah, it comes down to fixing, and it's the experience keeps evolving.
• 18:57
  SPEAKER_02: It's hard to fix it when it's always changing.
• 18:58
  SPEAKER_02: Yeah.
• 18:59
  SPEAKER_02: And you know, today's problems are are not necessarily tomorrow's problems, but as Ted mentioned, it always comes down to the person.
• 19:06
  SPEAKER_02: And you've got to spend time with that person.
• 19:08
  SPEAKER_02: You you gotta be with them.
• 19:10
  SPEAKER_02: It's it's that's just what it comes down to.
• 19:12
  SPEAKER_02: See what they're actually running into.
• 19:14
  SPEAKER_02: And I think those uh they did anthropic did those uh interview studies with their own employees and really talk to them and watch what they were doing.
• 19:21
  SPEAKER_02: That's super, super important.
• 19:23
  SPEAKER_01: I learned a lot from that.
• 19:24
  SPEAKER_01: Yeah, I love seeing that stuff.
• 19:25
  SPEAKER_01: I love seeing the the they're so good at collecting the data.
• 19:28
  SPEAKER_01: I love seeing the the trends on what people are using the tool for, what the what what what is like what are we doing with this tool?
• 19:36
  SPEAKER_01: Like how are we all using it?
• 19:37
  SPEAKER_01: It's always surprising.
• 19:38
  SPEAKER_02: Well, there's also the one guy saying that in in that article, uh the quote that it always that he's expected to do in one week what it takes other companies to do in like two months.
• 19:47
  SPEAKER_02: You know, so that's that's a lot of pressure.
• 19:49
  SPEAKER_02: And I think you I mean that's a whole conversation on its own, but it especially working at Anthropy, I can imagine it almost boxes you into like, how do I do the one week version of whatever I'm building right now?
• 20:00
  SPEAKER_02: And or the two month condensed to one week version of whatever I'm building, you know, you cut out that time that you would have taken to do a stroll around the office or to go out and talk to people to understand the problem better because you're you're expected to be, you know, yeah the powerhouse anthropic employees of a lot of pressure.
• 20:18
  SPEAKER_01: Is that is that the so if the if the cowork and the the chat interfaces are not the best UX, is it I mean, is are we reaching a point where it's like, yeah, we're faster at building things, but are we convinced we're building the right things?
• 20:35
  SPEAKER_01: Like, like, do you use codecs at all?
• 20:37
  SPEAKER_01: Do you guys use codecs?
• 20:38
  SPEAKER_01: Yeah, it feels like they just slam features in there in any order, in any, and if Sam Altman is listening, which I'm sure he is, I very much apologize.
• 20:47
  SPEAKER_01: But it's like, does it not feel like they just slam features in there in any order and then ship it?
• 20:52
  SPEAKER_01: Right and then they and then they deprecate it the next week.
• 20:55
  SPEAKER_02: Like it's it's tough, yeah.
• 20:57
  SPEAKER_02: It's tough, yeah.
• 20:58
  SPEAKER_02: Exactly.
• 20:59
  SPEAKER_02: You get attached to something, or you, you know, you the UI.
• 21:02
  SPEAKER_02: I mean you talked about the you made it like a reaction to the work versus chat mode.
• 21:06
  SPEAKER_02: And you're right, there's like there's things that one people love, one one one set of people love another set, you know, we just can't stand, and it's gonna it's gonna keep flipping and flopping.
• 21:14
  SPEAKER_02: We just need to decide one, just what just slow down, what are we gonna do?
• 21:18
  SPEAKER_02: Yeah, let's just decide on some stuff for a second, stop pushing it out.
• 21:21
  SPEAKER_01: I've said this before, Apple has got to step in and say, all right, guys, this is how you do it.
• 21:25
  SPEAKER_02: I mean, they're not even in the AI space, really, right?
• 21:27
  SPEAKER_02: So it's weird.
• 21:27
  SPEAKER_02: I think it's the weirdest thing.
• 21:29
  SPEAKER_02: Maybe I guess.
• 21:30
  SPEAKER_02: But it's I was I I would look to them to be like, oh, they this is how you do AI.
• 21:34
  SPEAKER_02: That's what I'm saying.
• 21:35
  SPEAKER_02: We've done for years.
• 21:36
  SPEAKER_02: That's what I'm saying.
• 21:36
  SPEAKER_02: This is how you do product design.
• 21:37
  SPEAKER_02: But yeah, you kind of can't do that.
• 21:39
  SPEAKER_02: You kind of can, but they're not like I don't know.
• 21:42
  SPEAKER_02: If they were producing papers like this, uh I would be really, really interested to see how their employees use AI and all that kind of stuff.
• 21:49
  SPEAKER_01: Yeah, well, awesome stuff.
• 21:51
  SPEAKER_01: Good stuff.
• 21:51
  SPEAKER_01: Well, thanks everybody.
• 21:52
  SPEAKER_01: Uh, that is it for this episode of Logixly Live.
• 21:56
  SPEAKER_01: Uh we are very uh thankful for Ashton for joining us.
• 22:01
  SPEAKER_01: We will Ted and I'll talk offline and we'll see if we invite him back.
• 22:04
  SPEAKER_01: I don't know if you're leaning one way or the other.
• 22:08
  SPEAKER_00: You wrapped up the road really well.
• 22:10
  SPEAKER_00: I was lost, I didn't know where that was going.
• 22:13
  SPEAKER_00: Good.
• 22:13
  SPEAKER_00: Because I was doing the math on 1945.
• 22:15
  SPEAKER_00: You said a thousand years, and I'm I'm like I was like, get you lost a little bit and then bring you back.
• 22:21
  SPEAKER_00: Then you brought me back.
• 22:22
  SPEAKER_00: I like yeah, that's awesome.
• 22:24
  SPEAKER_01: That's good stuff.
• 22:25
  SPEAKER_01: All right, everybody.
• 22:26
  SPEAKER_01: Um, as always, if if there's a topic you want us to cover, please reach out.
• 22:30
  SPEAKER_01: We'll be happy to engage and get it queued up for the next show.
• 22:34
  SPEAKER_01: And until next time, keep learning, even when it's unsupervised.