pri (
00:15)
we have a special guest, Michael, direct executive director of Rhizome. And Rhizome actually Hey Michael,
Aaron (
00:20)
Hey Michael, welcome.
pri (
00:22)
welcome.
Michael (
00:22)
Hi.
Michael (
00:23)
So nice to be here.
pri (
00:24)
great to have you. You guys are also celebrating
Michael (
00:25)
On the hundredth.
pri (
00:26)
a milestone this year as well.
Michael (
00:28)
We
Michael (
00:28)
are yeah, just our thirtieth anniversary. It's time Rhizome
Aaron (
00:32)
Let's go.
pri (
00:32)
That that's huge.
Michael (
00:34)
is grown up and ready to move out of her mom's basement.
pri (
00:37)
Yeah.
Aaron (
00:37)
Barely.
Aaron (
00:38)
Barely.
Chris (
00:39)
You gonna get a porch? A sublet?
Michael (
00:39)
Just about. Getting there. Within five years.
pri (
00:39)
Thirty parents?
Chris (
00:42)
Derek, since we're celebrating being washed here, do you have any unk milestones to check in with?
derek edws (
00:48)
Man, I mean I turned I turned forty years old this not too long ago, just a couple of months ago. I don't even know if I shared that here on the pod, but I'm I'm getting up there. that is
Chris (
00:57)
That's a milestone. Congrats, man.
derek edws (
00:59)
thank you. I appreciate that.
Aaron (
01:00)
We're gonna live to be like two hundred, Derek. You're still a baby.
derek edws (
01:04)
I appreciate that guys. Yeah, definitely the time is speeding up, it feels like. I you know, I just got back from New York. I was out there for
derek edws (
01:10)
I got a chance to see you guys. We we got to record an episode and I ended up staying out there for a little bit longer just to do work stuff. And I I did get a chance to go to Rhizome's annual gala. and I I admittedly, you know, as someone who's been collecting digital art for the better part of the last five, six years, my interactions with Rhizome has have been kind of few and far between. And so this was maybe the
derek edws (
01:33)
The most time I had spent thinking about and understanding and learning about Rhizome. And I thought the event was incredible. maybe for the benefit of folks who who haven't had a the chance to understand the the great work you guys are doing, Michael, maybe you can just give a quick spiel on on what you guys do and and how long you've been doing it and maybe the mission or the focus of what the organization looks like right now.
Michael (
01:56)
Yeah, sure thing. well, Ryze you know, we have a kind of fascinating history. We were started by an artist named Mark Tribe in ninety-six as a mailing list. so he didn't really know where it was going, but he just knew that he was seeing s the same people at, know, similar kinds of events, and he really saw this network that was forming around new media art that was really gaining an energy with like the rise of the internet.
Michael (
02:21)
So we started an email list and it was a place where people could share their work and kind of develop a shared language around what digital art could be and share resources and stuff. And it just kind of took on its own momentum. he ended up coming, he was in Europe, then he came back to New York, started a dot-com. Ryzen was gonna sell digital art to corporate websites.
Michael (
02:43)
And then you know, started kind of a magazine aspect through parties and eventually and already within three years, realized that there was a, you know, kind of pressing need for an archive of this work that it was really hard to maintain over time. So in ninety nine, Mark and the Rhizome team launched an archive called the Art Base.
Michael (
03:05)
Which was a kind of catalog of new media works, some of which were hosted by Rhizome and some which of which were hosted elsewhere. And that really kind of defined what like our unique role in the world since then in many ways. Today I I describe Rhizome as a memory institution for digital art. we do a lot of technical and curatorial research around how to maintain access to old digital works. Sometimes we work in partnership with institutions,
Michael (
03:33)
like a long we have a long running partnership with the Walker Arts Center, maintaining their digital collection. but we also kind of haven't lost our sort of community and experimental edge and we, you know, we support artists through projects like our micro grants and which are actually coming up. People can apply to get a small grant to make an online work and
Michael (
03:54)
You know, we have events and we also have our program Seven on Seven, where we have artists and technologists collaborate to make new things. So we kind of, you know, really haven't lost our commitment to like experimentation and and you know, staying with community. and I guess like one historical note is that as part of seven on seven in twenty fourteen, that's where Kevin McCoy, the artist, minted quantum, which is kind of a
Michael (
04:21)
Important historical reference point in the early prehistory of NFTs. But yeah, I mean, I I guess that that's sort of how I think of us today is memory institution plus experimental community. And we are online organization. So, you know, our events pop up in different places, but we do have a longtime home base at the New Museum in New York, which is on Bowery.
Michael (
04:44)
and recently reopened. So a lot of our in-person events do happen there, but we you know we work online and with the museum and with other partners too. So kind of a distributed model to some degree.
pri (
04:55)
Yeah, I
pri (
04:56)
mean you guys have been doing great,
Aaron (
04:56)
Yeah, that was a great summary.
pri (
04:57)
great work. and I forgot that Kevin minted quantum at seven on seven.
Michael (
05:03)
Yeah.
pri (
05:04)
that's yeah, that's it's obviously you guys have had like just so many artists that have come through and like also digital art. If you guys were so early being thirty years old, like I just feel like the entire space has changed so much probably in those thirty years and it really is like the art of your our our generation and time and so
pri (
05:23)
Yeah, it's amazing. I mean there's no way that like like I I feel like every major artist has had something with rhizome. and so it's amazing to see and congrats on thirty years too. It's awesome.
Michael (
05:33)
Thank you.
Michael (
05:34)
Yeah, the Guinness Book put that Kevin McCoy NFT into into the world records for yeah.
pri (
05:39)
They did.
Aaron (
05:41)
Let's go.
Michael (
05:41)
But it's so it's a
Michael (
05:42)
source of great frustration for me as an Irish an Irish person because they didn't mention Rhizome. And I so I like wrote to the Guinness book like, Come on, guys, this was at a rhizome event. You gotta put us in there if you're gonna if you're gonna do this. And they were so mean about it. my people are my
Aaron (
05:57)
Really? Why?
Michael (
05:58)
my people are forsaking me.
Aaron (
06:00)
That's surprising. What
Michael (
06:00)
So we're not listed. Yeah.
Aaron (
06:02)
do you think the next thirty years looks like? Do you think you turn those memories into like a like a like an AI model of some sort? Have you ever thought about stuff like that?
Michael (
06:12)
I mean,
pri (
06:13)
Yeah.
Michael (
06:14)
our memories have been turned into AI models. it's actually really crazy how much energy we now spend on keeping our servers alive even while AI bots are sc scraping them constantly. so yeah, the data is sort of like I mean, I think about this a lot. You know, this first thirty years, it is kind of like an artisanal data set now.
Michael (
06:36)
You know, this is all information that was assembled, you know, pretty much by hand in the pre AI days. I feel like that's kind of like gold in the data onslaught of the future. This kind of, you know, like
Aaron (
06:50)
It's the only thing that's gonna
Aaron (
06:51)
matter matters to the Pope, right? So there's some significance I guess being ascribed to it.
Michael (
06:54)
Yeah. But I think it
Michael (
06:55)
actually matters to the people training models also, you know. Like he's making an ontological claim, but the people that are making models are like, my gosh, there's a server from two thousand and three that's containerized. We want that. so there's kind of a constant, you know, constant traffic of of training happening. But I think like for our own purposes, definitely interested in in things that we can do to like
Michael (
07:21)
you know, maybe create other kinds of interfaces or discovery of, you know, what's really a overwhelming amount of data at this point.
derek edws (
07:30)
Out of curiosity, Michael, I think this is like a maybe a good segue. The the annual benefit that you guys had were celebrated they kind of like honored two different artists. One was Scabanetti and Scabonati, am I pronouncing that correctly?
Michael (
07:43)
Going out of yeah. Mm-hmm.
derek edws (
07:45)
And and then Snowfro, who's a blockchain based artist. And I I guess I'm curious to hear how you think about yeah, the role of the memory keeper in a world where
derek edws (
07:56)
some or many of the works that are now intersecting with digital have this public database that they're kind of writing the work to and where you see Rhizome kind of both playing a role and and where play where where best to play a role in a world where maybe blockchains are, you know, becoming more and more used as a tool for recording digital artworks.
Michael (
08:22)
you know, Eric and I have conversations about this that are kind of interesting, but I th I think that you know, there's just endless examples of the ways in which like
Michael (
08:31)
narration and trust do matter in claims of provenance or cultural value. and you know, on a kind of cure curation level.
Michael (
08:42)
The the thing that differentiates like a valuable work from a non-valuable work, like, you know, and there's only certain aspects of it that can I think be captured cryptographically, you know. like what the work meant to a community or something. Is does all of that show up in the in the blockchain? I'm not sure. so I I think that I certainly haven't seen like less of a need for like
Michael (
09:08)
you know, institutions that like that like continue to like tell stories or kind of create moments around, create experiential kind of like encounters with that kind of like rehearse like forgotten things, like kind of go through the data that's like, you know, whether it's on blockchains or old servers, going through that material and and kind of creating meaning from it is still or maybe even more of a pressing task than it used to be.
Michael (
09:36)
and you know, the even on a technical level, this is something that Eric where Eric and I have like ongoing conversation, but you know, if you think about a work that's running from chain that's stored on or sorry, from code that's stored on chain, it still needs to be executed in some kind of environment. And there are a lot of ways in which like Rhizome's technical preservation work is focused on
Michael (
10:01)
software environments that are preserved, not the underlying data. So the problem that the blockchain has solved, which is the problem of storage, is is the one that we work on the least. the one that we work on the most is cataloging and understanding software. And there's a lot of interesting stuff that can be done with that now, like using AI tools. And we're, you know, we're in the kind of
Michael (
10:24)
process of like researching all that. So I think, you know, curatorially, meaning making, experiential encounters, and then even like software environment preservation, none of that stuff is like really solved by the blockchain. It just sort of becomes like I think another context that we intersect with.
Aaron (
10:42)
Do you think that becomes the well more and more work is, Michael? Like looking at like all these like cracked math problems.
Michael (
10:50)
huh.
Aaron (
10:51)
you know, it's kinda like the the job of like these expert mathematicians, and we pre and I were looking this up yesterday. I think there's only like several thousand in the US that that do this, and only like several thousand more that do that globally. It's more like understanding the
pri (
11:05)
The auditing.
Aaron (
11:06)
context, yeah, auditing it, you know, trying to make sense of it.
Aaron (
11:11)
'Cause it it feels like that type of a task, almost like the archiver, you know, soothsayer, Oracle of Delphi type approach is is
Michael (
11:21)
Yeah.
Aaron (
11:21)
where we're where we're heading.
Michael (
11:23)
Yeah, I mean there's a lot of I think what's interesting is there's a lot more people kind of doing this work than there used to be. And that's something that I I always anticipated that to be like a positive outcome of the NFT boom. And it's it's really I think borne out that there's so many more people kind of like looking at looking at the history. And I feel like in digital art and culture it's actually so hard to retain any sense of what came before. Like
Michael (
11:51)
You know, what did your iPhone interface look like five years ago? Like what when you wrote a a blog post in a certain way, like what how did the nature of your site at that time change the way that you wrote it or constructed it? That stuff is actually really hard to hold on to. I think because so much of it feels like mundane or everyday or incidental. And and so as we kind of move through our digital society.
Michael (
12:18)
like understanding how we got here is really hard to do. It's a technical problem. It's also kind of like a you know, like something about our human memories in the way that we interface with these devices. And
Michael (
12:31)
So the people that are thinking historically are really thinking about how the past can be a resource for the present in the future. How, like something crazy that someone did five years ago, an experiment that didn't work then, you know, maybe revisiting that or thinking about what was, you know, what the next step could be. Putting together those histories is really important. And it's really magical like to have the experience of like looking at some kind of
Michael (
12:57)
resource. It's been sitting there in plain sight for years and years, whether it's a blockchain or the Wayback Machine or even just someone's website. And then just coming across something and thinking, wow, that's actually this really important thing that no one has talked about in 20 years, you know? it's like an incredible experience. And I think there's a lot of p kind of historians out there, like not just RISO, but many others that are doing that kind of work. And I think that's a really interesting aspect of this moment.
Michael (
13:25)
it's kind of yeah, kind of like what you're saying, like there's the cryptographic wizards and then there's the people that have to sift through and interpret.
Aaron (
13:33)
It feels like work's gonna be a lot more like sifting through things. Like if there's a hundred years of mathematical innovation every month, right? Or maybe every week, like how you gonna keep up with that? Feels like you're gonna need
pri (
13:46)
I mean
Aaron (
13:46)
a bunch more mathematicians that are sitting there kind of doing that archiving, you know, like
pri (
13:50)
My one Yeah.
Aaron (
13:52)
knowledge synthesis pieces.
pri (
13:54)
My one thing with that though is like, what if we can't understand exactly what they're doing and therefore people just stopped caring? Because it's like somehow like you're getting to the answers, we just actually don't know how it's getting done. We can't decipher it because it's like in a language that we don't know. it's like in a gibberish like, is it worth the human brain power and effort to like understand how it got to a conclusion?
Michael (
14:17)
I mean, are you I it's really crazy to think about a field like computer science where if you came up in the 70s, you you would know the whole field of computer science. Like one person could study computer science and know all of it. And like today the specialisms are getting so crazy. So let's I I think you're I you're kinda onto something that like none of us can really
Michael (
14:39)
It's like really hard to synthesize these extremely specialized aspects.
Aaron (
14:44)
Well and also I don't know if you took a look at the output from the math, like problem breaking, but it's it's almost like inscrutable like how they're what they're saying, like their reasoning. I think it's gonna get like a little a little bit more religious. Like people won't even understand how they got certain answers or certain things, certain solutions solved. So
pri (
15:07)
I mean
Chris (
15:07)
Is that
Chris (
15:08)
really a problem? I mean on a
Aaron (
15:10)
I don't know.
Chris (
15:11)
On a practical basis, there's certain things you absolutely have to understand because the edge conditions or when they don't behave as expected can have catastrophic or, you know, have dangerous implications. I think there's a whole other range of stuff in which you can just abstract all of this away. And because the way you're framing this conversation, it's almost like you're putting yourself downstream as a machines.
Chris (
15:38)
And that that just doesn't sit well with me. Like you should be on top of the machines and the machine should work for you. And so if you're really just repositioning this whole world as, AI spits all this out and then we have to sort through the you know, the great garbage pile and f pick pick pick things out and be fascinated by them, like that's just a a very positioning. I it just doesn't fly for me.
Aaron (
16:04)
Pre
Aaron (
16:05)
pre modern type positioning, Chris? I mean we've we've done that for a while.
Chris (
16:08)
Yeah, it's very beta,
Chris (
16:09)
it's very like l loss of agency. I'm just not feeling
pri (
16:12)
I kind of agree.
Chris (
16:13)
this.
pri (
16:13)
Yeah, I kind of agree. It feels a little cuff.
Aaron (
16:14)
But I think we I mean most
Aaron (
16:15)
of it it may be cucked, but it's how most how humans operated
pri (
16:19)
Just
Aaron (
16:20)
through most of history, right? Like we've only been able able to understand our world, whatever s well, servants of the
pri (
16:24)
Just servants of the machine? I don't think so.
Aaron (
16:27)
the greater the greater world, you know, our greater environment.
Aaron (
16:32)
Like if we were way back in ancient Greece, like
Michael (
16:33)
Well, that's a different bit of thing though.
Aaron (
16:34)
they w they didn't understand anything either. You know, they were just subservient to the whims of you know, weather,
pri (
16:39)
That's true.
Aaron (
16:40)
disease, everything else like that.
Chris (
16:42)
Okay. A, we don't really know how like people people's worldviews were back then. You you can you know try to make sense of it via what has survived. And that can create all sorts of different biases. But also those people, because now I'm gonna do it myself, those people were very much like in touch with their environment. They you know it it's not like they were sitting and staring at screens.
Chris (
17:09)
They were experiencing phenomena that they tried that they had to like apply some sense making to. And the sense making that they applied to it was like the creation of this, you know, whole whole like mythological cosmology. And so, you know, it's the will of the gods. Well, they invented the gods, right? And so it's not necessarily, you know, if they were downstream, you know, floundering about. You know, I'm just gonna I'm gonna keep pushing back here.
Aaron (
17:37)
Yeah, I I hear it, but I I think that may we may push back, but I think that may be the direction we're heading.
Chris (
17:43)
No, not not not taking this one. Not going gentle into this good night. The machines work for us.
pri (
17:47)
I also feel like
Aaron (
17:48)
You're r you're
Aaron (
17:49)
this is why we all loved Rage Against the Machine, right? They were just telling us what we were gonna need to do.
pri (
17:54)
Yeah, and like Chris, does it worry that you might actually not be able to decipher what they're saying about you behind your back?
Chris (
18:00)
No.
Aaron (
18:01)
shit, they're
Chris (
18:01)
No it doesn't.
Aaron (
18:02)
pre, they're only gonna be talking about you behind your back, not me and Chris.
pri (
18:05)
Make it about me again.
Chris (
18:07)
I well, I I'm not the one like thinking machines are gonna gossip. I'm not the one anthropomorphizing them. Like these things are built to perform tasks.
Aaron (
18:13)
they d they're definitely gonna gossip.
pri (
18:13)
I am.
pri (
18:16)
Dude, are you kidding? They like
Aaron (
18:16)
No, dude.
pri (
18:17)
the first thing they do is build a a board to talk to each other. They're like humans. They humans
Aaron (
18:20)
to shit post, yeah.
Michael (
18:21)
Yeah.
pri (
18:22)
love gossip.
Michael (
18:24)
There does seem some like a fundamental difference between the deterministic kind of computation of the past and what you know, what these AI agents are doing because they're not following a task. You set a task and then they've they follow it, but they also have like another latent set of goals or priorities that somehow comes from training or some you know, or from
Michael (
18:51)
like it's really hard to kind of like truly align AI, you know, in and say, like, I guess that's what you're saying. It's like when you give an A AI a task and you expect it to complete it, that is a reflection of whether that AI is sort of aligned to, you know, to your goals. And yeah, it seems like it seems like the jury is very much out on how that's really gonna work at this point, even while
Michael (
19:18)
you know, these things are deployed in all kinds of critical ways already. so I think alignment you know, people often always like cast it as like, you know, AI sending a nuclear bomb to wipe out the earth. But I think on a like more mundane level, it's it's also just like, you know, you you set a an agenda, you know, does does the agent follow the agenda or is it operating according to some other principles?
Aaron (
19:44)
Yeah, as Chris knows, they're they don't have great executive function. I think like we're gonna see like new academic fields like agentic organizational design, or like new majors in like building intelligence factories, s stuff like that. It it feels like it's gonna be its own discipline of study.
Chris (
20:04)
You can go to school for network coordination studies.
Aaron (
20:07)
Yeah, exactly, Chris. Yeah. You're gonna go through you're you're gonna get you know, your AI black art black arts class, a little bit like in Harry Potter, where you learn how to read through the tea leaves of the AI system and then you know, shuttle down the hall to your agentic organizational design class.
pri (
20:25)
What
pri (
20:25)
what department is this under? Is this under so like social sciences? Like w w what what do we what what what department is this a whole new wing? Is this computer science? Like where is this sitting so yeah. Where is this sitting in the academic
Aaron (
20:35)
I mean, you could argue that it Yes, the cybernetics depart department.
Chris (
20:36)
Kubernetes.
pri (
20:40)
level?
Aaron (
20:41)
Yeah, I mean pri computer
Michael (
20:42)
Mm-hmm.
Aaron (
20:43)
science, right? I mean if computer science solved every other science, then it kinda becomes the alpha predator science, right?
Michael (
20:49)
No.
Chris (
20:50)
I don't know, Aaron, that's like saying paper would become the the dominant
Aaron (
20:54)
I mean yeah, but I mean if
pri (
20:54)
Yeah, that's weird.
Aaron (
20:55)
the study of compu computer science led to the you know, the muting of almost every other science. It ha it it's kinda the only one remaining.
Chris (
21:03)
I think that's all hyperbolic, man.
Aaron (
21:05)
I'm I'm just stirring the pot, don't worry.
Michael (
21:06)
But I I think it's
Michael (
21:07)
when I think about your question, Chris, it's sort of like, you know, or the the kind of response you were making. Like let's say there was some example. I think it was a little bit of w one of these fake news stories about the an an AI bot like was asked to perform a task on email and discovered a crime or something and shared information about it. Like I don't know if it actually happened or not.
Michael (
21:30)
it was just reported in the news, which is not a good indicator of underlying reality these days. But I think like in an example where there's a you know, there's an order that's given to the AI to complete a task, something else happens, you know, then it becomes like, why did the something else happen? Was it because the AI entity had some anthropomorphic quality that we're ascribing to it? Or is it just like some other kind of like statistical computational noise? And I think
Michael (
21:58)
It you know, I think the reason it doesn't the question doesn't belong to philosophy and social science is that it's you know it isn't it isn't a you know a human-like entity, I don't think. this is an article of faith now among people in the field, but you know, this is a kind of complex mathematical system that then leads to certain outcomes. And then they look like those outcomes look like, you know, human decision making. They look like human wishes.
Michael (
22:26)
Yeah, but
Aaron (
22:26)
I mean not according
Aaron (
22:27)
to Anthropic now, right? You can't you can't yell at your at your claude. They want some decorum in your
Michael (
22:32)
Well, I feel like it's
Aaron (
22:33)
in your chat rooms.
Michael (
22:35)
It's just a bad habit to yell at things probably. But, you know, if you're yelling at your
Aaron (
22:40)
I mean I don't disagree.
pri (
22:41)
Sometimes I can't sometimes
pri (
22:43)
I can't help it.
Michael (
22:44)
What's that?
Chris (
22:45)
sometimes they have it coming. And
pri (
22:47)
I'm dying.
Chris (
22:49)
You know, the
Chris (
22:49)
people frame all of this stuff so very much like around the AI itself. And then they want to add other properties around it, but like i what you just brought up, Michael, I could say just as much is influenced by private law. And that private law extends everywhere, and that your use of services comes with a whole set of
Chris (
23:14)
Assumptions, rights or lack of rights. And that it would be, you know, in your example of like an AI found a crime in an email, what does it do? So much of that, I think, is actually built around like the service provider's private sphere of law and how it already behaves and how it operates.
Chris (
23:36)
And that's just passed down into the AI itself. And so people want to frame these things as like ethical quandaries within AI when in fact this is really more about like, you know, how we've given up certain rights for access to tools that make our life easier. And that this is just an extension of that. And you know, to like frame it entirely within the realm of, my God, this AI is just like a little person and it's like a little judge and it's making a moral decision.
Chris (
24:03)
Like that's fine. Like that's a perfectly valid framing, but it also discounts what put it into that situation. What you know, what of its training has it like debating this even as a thing to begin with?
Michael (
24:16)
Sure, but if if you trained a human operator and the human operator saw the email and passed it to law enforcement, then you could you know, then the law could say, actually that's not okay. Train the operator not to do that. And the I think what's interesting
Chris (
24:29)
Well the law would
Chris (
24:30)
say no, that is okay. And thank you for eroding Fourth Amendment privileges. We we've seen this happen in Geek Squad. You know, like people take their computer into Best Buy and then they're getting arrested and
Michael (
24:41)
Well, but
Michael (
24:42)
that's not always okay. It's not always okay to like look at consumer data and it it also depends on the crime, you know. So I think what I'm trying to say is that like the my understanding is that what the AI agents behavior isn't like legislatable in the same way.
Michael (
25:00)
Like it's based on training patterns and it's understanding of a situation, like or it's or or like some kind of statistical analysis of a situation. But it's more difficult to be like, you know, the the law has changed, now do things this way.
Michael (
25:14)
You know, it's like because it's like not the deterministic task based completion of you know, of traditional computation.
Chris (
25:22)
Right. It has reasoning capabilities, and those reason capabilities are maybe bounded by what their training set was and you know what the model weights were, and therefore it's both capable and incapable, right? Depending on how it's actually tuned and configured. to me though, that like isn't necessarily a function of like that AI itself. It's a function of like what built it.
Chris (
25:46)
And you know, kind of picking hairs here, but like I just I always want to push back against like imbuing too much into AI when AI is really like a distillation of a lot of different societal behaviors, norms, what you know, like it's training on what's already come before. These things, you know, aren't necessarily like in net new
Chris (
26:12)
you know, self recursive, self discovery training based on their own experiences out in the world, they're they're more like, you know, stochastic parrot still.
Michael (
26:22)
But it's interesting 'cause the whole term agency kind of relates to like structure and agency problem, where agency is the part of you that doesn't get to be determined by the system. But what you're kind of saying is like the AI is totally determined by the system that produced it.
Chris (
26:37)
At this point in time it largely is. Right. We we haven't evolved them to the point where I think that maybe some of the debates we're having, right, are gonna be applicable three years, five years, ten years from now. We're just jumping the gun a little bit because, you know, we want to we wanna see ourselves in everything.
Michael (
26:58)
Well, we are pretty cool, so
pri (
27:00)
Well. We have to make it better ourselves.
Chris (
27:01)
I don't know, you ever seen a squirrel wander around, dig up nuts?
Aaron (
27:05)
Yeah.
Michael (
27:06)
Actually that sounds pretty good right now. I'm gonna go find a squirrel. Yeah,
pri (
27:08)
Just look watching squirrels at the park.
Michael (
27:11)
we have a beautiful day in New York for squirrel watching.
pri (
27:13)
I know, that's true, that's true.
Aaron (
27:14)
That's true.
Aaron (
27:15)
Maybe that's all we'll need to do, in the future.
pri (
27:17)
Actually sidebar pigeon
pri (
27:19)
pigeon pigeons are cool with Gen Z. I don't know if you knew that, but pigeons are ha are making a huge comeback. I swear, I swear. I read it I read
Aaron (
27:25)
What what are you talking about, Pray? What do you s what what what does your algorithm do to you, dude?
pri (
27:32)
an article about this, like literally this week.
pri (
27:35)
Like i it it's the new Gen Z hero is a pigeon. They like they are like adding it to fashion trend, like JW Anderson has done that. There's also been so there's like fashion nods to it, but people are also doing a lot of like pigeon watching and they're also sending messages via pigeon too. It's like a whole thing.
pri (
27:57)
Anyways, we don't need to we birding in general is back with Gen Z, but like pigeons as a symbol are cool.
pri (
28:05)
Well, I got it.
Chris (
28:05)
This
Chris (
28:05)
is totally Mo Williams' fault. Everyone who read all his pigeon books as kids have grown up
Aaron (
28:08)
Yeah, completely. They got incepted.
Michael (
28:09)
Yeah.
Chris (
28:11)
now. Don't let the pigeon
Michael (
28:12)
I'm trying to think of my
Chris (
28:13)
drive the bus.
Michael (
28:15)
I think my kids are Gen Alpha, but they are actually also into pigeons. But mostly they they really hate bugs though, which I like I feel like I spent a lot of my childhood playing with bugs.
pri (
28:27)
me too. But you know what? I
Michael (
28:29)
And are there like less bugs now?
Chris (
28:32)
There's less
Chris (
28:32)
backyards, there's less kids out in backyards digging stuff up.
pri (
28:36)
I used dig up worms. It was pretty gross. I had the insect club.
Chris (
28:41)
Yeah,
Chris (
28:41)
we used to go snake catching in my neighborhood.
Michael (
28:43)
wow.
pri (
28:44)
That's that's dangerous. Garden snakes. Yeah, yeah. Those are fine. I
Chris (
28:47)
No, 'cause they were like eastern garter snakes. It wasn't like in that we're not in Texas.
pri (
28:51)
have like a core
Michael (
28:51)
Please picturing another.
pri (
28:53)
Yeah, I know. Like where I actually have a core memory of like like there was like a a a snake, a garden snake in my garden. My mom like tried to like take a shovel to scoop it up and like put it somewhere else, but then s like sliced it in half, like literally sliced it in half and both sides were just like wiggling around forever and it was just like disgusting and weird.
pri (
29:12)
to like see both sides just like move for that much for the yeah, it was it was disturbing. Anyways, I don't need to like traumatize you guys. on this.
Michael (
29:20)
I'm gonna have to check out Pigeon TikTok later.
pri (
29:22)
Yeah, it's a whole thing. It's a whole thing.
Chris (
29:24)
It's a great thing to do when you're in line. Stand in line, watch pigeons.
pri (
29:28)
Ugh, God, are you standing in line now?
Chris (
29:31)
No, I'm pacing around my basement. I don't stand in lines.
pri (
29:34)
get your Mimi's frozen yogurt.
Michael (
29:36)
Ha ha ha.
pri (
29:37)
Well look at some pigeons. It's very Gen Z of you, Chris.
Michael (
29:40)
Have you guys done an episode on
Chris (
29:40)
I'll feed it I'll feed the
Michael (
29:41)
the
Chris (
29:41)
rest
Chris (
29:42)
of it the squirrels.
pri (
29:43)
Yeah, and and look at squirrels. I guess just to go back to the AI conversation that we were having though, as as AI starts taking our work and we start, you know, getting back into our roots of like pigeon watching and doing things like that. something that we were talking about at the beginning of the call, like just to tee up what we'll talk about, was just like hardware coming back a little bit. And I know you guys talked a little bit about this on the prior episode, but like
pri (
30:11)
garage tech and and people starting to like want to actually start building hardware. something that's things that feel tangible. Obviously all of us here love the the digital realm, but it does feel like hardware plus digital is making a comeback. And we were talking about Snowfro too when it comes to this. But like Michael, do you see a lot of
pri (
30:35)
the digital artists like playing around with hardware now more than before.
Michael (
30:39)
Yeah, a hundred percent. I was thinking a lot over the last year about this phrase software is going to zero, you know, because software's gotten so much cheaper to write. This is what SARM
pri (
30:49)
Yeah.
Michael (
30:50)
from IDEM IDM is always saying. Software's going to zero. So the corollary is that hardware is going to a hundred or so people are definitely really into hardware again.
Michael (
31:01)
And I think that it's a reaction partly just to the I mean, there's an economics aspect of it because suddenly your old computer is actually worth money, which is a really weird situation. like the thing that would have been just like off to the e-waste bin five years ago is suddenly like two hundred bucks, you know? And so people are spending a lot more energy
Michael (
31:22)
like setting up a net BSD server or figuring out stuff that they can do with their old machines. so there's like an economics aspect of like why people are, I think, picking up like getting into like the guts of hardware again. And then, you know, there's been this whole cyberdeck trend that people are aware of and the perma computing movement, cottage core tech. and
Michael (
31:43)
I'm curious actually, like I think part of this is like people running local AI models, which are very you have to kind of get into the hardware scene a little to understand too. so all of that is kind of like the general backdrop. And then within that, you know, I think artists are there's a lot of like people that are working on devices and things, but I think part of it is also just when you're an artist.
Michael (
32:06)
Making work that fits inside like an iPhone screen or an Android screen and then like is just surrounded by notification swarms is not fun. So having another thing that you can put your work on is like a way out of the attention economy madness a little bit. so people have been I've seen a lot of experiments in this realm and
Michael (
32:27)
I think we're probably at the early stages of it, but I I feel like this year there'll be a lot of artist hardware experiments that we'll be seeing floating around.
Michael (
32:36)
I'm curious if are you are any of y'all running your own AI models at home or have your own server in a closet?
Aaron (
32:43)
I haven't yet. but I feel like next year that's gonna be a big trend. I I don't think they're good enough quite yet, but I think the moment that they are, they'll get pretty widespread use because you know, still to do if you wanna build like really like a persistent agent of some sort, it's pricey. And so I think they're gonna need to get localized
Michael (
33:02)
Yeah.
Aaron (
33:02)
at some point. you know, if we could get
Aaron (
33:05)
something that's like running in Grokbot or Muse or Instinct, but just run locally, so it cost you effectively just your hardware cost and electricity. I think you'd see really an explosion of different use cases. I think still a number of use cases are just too expensive, especially around like media media
Michael (
33:24)
Yeah, for sure.
Aaron (
33:25)
stuff.
Michael (
33:25)
I was looking at a budget for setting up a Kimi the latest Kimi model running locally and it was it was about sixty thousand for all the hardware. So it's not cheap.
Aaron (
33:35)
R yeah, that's
Aaron (
33:36)
a completely unattainable, right? So but I
Michael (
33:39)
Yeah. But I think that the
pri (
33:39)
Yeah.
Michael (
33:41)
the thing that's interesting though is that like there's a lot of old models kicking around that still have like, you know, artistic value and stuff. And Dan Keller actually has been working on this. Have you guys had him on the podcast here? I think he's connected to this crew. Have you seen his old models website?
Aaron (
33:55)
Yeah. We work with Daniel on Aiden. Yeah.
pri (
33:57)
Yeah.
pri (
34:00)
Yeah, it's
Aaron (
34:00)
Yeah, we've been
pri (
34:01)
cool.
Aaron (
34:01)
super supportive of that.
Aaron (
34:03)
we were that was part of the collecting thesis too around in Flamingo for some of the early AI works. Like as the like march towards superintelligence continues, like the the original outputs would be something people would find interesting or
Aaron (
34:19)
Or wanna celebrate or, you know, study, all that jazz. So it looks like that thesis is beginning to play out. Like if you look back at like some of those earlier, you know, image or video models, I mean they're they look insane. Like absolutely insane.
Michael (
34:33)
Yeah, totally.
Michael (
34:35)
They were so funny. They were so great. We really lived in a golden age.
Aaron (
34:38)
Yeah, like the
Aaron (
34:39)
like limited color set that they seem to pull from and the like trippy psychedelic like motifs that seem to be coming out of them to start it's kinda wild. It's like the
Michael (
34:49)
Yeah.
Aaron (
34:49)
birth of you know, some collective con collective consciousness. no, Chris, did I go there?
Michael (
34:55)
Yeah.
Chris (
34:56)
Yeah. No, th I think there was a
Chris (
34:57)
lot interesting about
Michael (
34:58)
Ha ha.
Chris (
34:59)
old models. you know, I you my prompting style was very, very different with old models because I think it was more oriented around exploration or it was more feedback driven based on what the model was capable of doing. I would try to steer into those abilities.
Chris (
35:20)
You know, versus nowadays when like the fidelity is so much better. You're very like mission driven or focus driven in terms of like I want this particular output in this way. And those are two very different orientations, for sure. And that that sense of play or exploration that existed in old models, you know, we we certainly have of lost something.
Chris (
35:44)
You know, people talk about the image and video a lot. For me, that was very much like I I was very deep into Suno for like a year or so, right? I just loved making music and it was like a new expression of of writing for me, like actual songwriting. And it was very like a bite-sized way to make a thing, you know, that you could do like during like a a half of a football game. And with within Suno, there was definitely a tipping point, like around like version three five or so, three five to four.
Chris (
36:13)
Right, when that sort of like emergent properties of wow, you know, this set of you know styles when combined together within a prompt can produce this sort of sound that I've just never really heard before. Let's see what we could do with it. And you know, as they got better and as they got, you know, more I don't know, consumer grade.
Chris (
36:38)
Right, you lost that like bizarre steerability and and half the fun went out of it. You know, like I haven't touched Suno in God knows how long, y you know.
Aaron (
36:46)
I actually was playing around with it the other day. It it's definitely better. I don't think it's like leagues better than when you were playing with it though. Playing with it though, Chris.
Chris (
36:54)
Yeah,
Aaron (
36:54)
It's like a l a little bit more
Aaron (
36:55)
steerable, like a little bit cleaner.
Chris (
36:57)
Like maybe I'll get back to it, but you never know. But like old models, man. Old models are great.
Michael (
37:03)
Yeah. The only model I run locally is Whisper, I guess, which kind of barely counts, but I love it. I you know, like 'cause we always end up with like some hour long video conversation of something that I have to like sift through. And now I can transcribe it on my computer for no money and without sending it to the cloud. And it's beautiful.
pri (
37:24)
Wait, on whisper?
Michael (
37:26)
Yeah, Whisper is like a local open source AI model for voice to text.
pri (
37:31)
I I I use Whisper, I mean I pay for it, but I didn't know it was like local.
Michael (
37:35)
Yeah, you can run it locally on your you you are you using Whisperflow?
pri (
37:39)
Yeah.
Michael (
37:40)
Yeah, that's I think based on the open source Whisper model that you can run locally.
pri (
37:44)
I didn't realize that.
Michael (
37:46)
Yeah, but I mean I I think to do the nice interface and workflows around it, you would have to like do a lot of scripting. So I'm just like plopping video into it, but it's really I really like it.
Michael (
37:57)
it makes me think there could be like really specific models for other things that are manageable in si in size, you know?
Chris (
38:05)
Yeah, I mean that's an interesting little like corner of the world because there's certainly a motivation, right, to be able to run models on browsers, to r to run models locally.
Chris (
38:15)
There's there's definitely been a few where, you know, they can get down to like eight gigabyte size or like there's extremely tiny LLMs that you could actually embed in the browser, right? but there's still like a gig and it's not something you just want to like slap into an application. And so when can we start getting more and more of these microservice models and you know, what will they allow for is is certainly interesting. Like you
Chris (
38:42)
You know, you can run like zero shot classifiers pretty easily, you know, in applications, like they're teeny tiny, but like they're limited in what they can do. But then you look at something like the JEV you know, system one approach to to models that are very fast and answer all sorts of small little questions. you know, like that's a whole new design pattern and and and that opens certain things up. And and so little mo little models, just like old models, they got their place, they got their purpose. We just
Chris (
39:10)
you know, need to wait for a few more of them and need to figure out how they all, you know, play in this world. But it it's cool stuff.
Michael (
39:18)
Yeah. Yeah. I mean, I guess kind of like with the cyberdeck conversation, a lot of the way that artists initially kind of picked that up was like very decorative, you know, making a computer that had like a certain kind of case. But I'm kind of more interested in it what I think will happen now, which is more about the guts and you know, trying to s sort of like make like an exploded computer to run a small model for some purpose or some kind of obscure computational thing.
Michael (
39:45)
But you Eric's project is kind of in a third category, which is a like a like a bespoke kind of artist's device. And do you guys have it actually?
Aaron (
39:54)
Yeah, I think
Aaron (
39:56)
Yeah, we do. I think it it had a couple
pri (
39:57)
Yeah.
Aaron (
39:58)
it had a couple issues, but it it's definitely interesting.
Michael (
40:02)
Yeah.
pri (
40:03)
Yeah, it's super fun.
Aaron (
40:04)
but
Aaron (
40:05)
I do think I mean that's kinda I like I saw somebody they built like a Tamagotchi breathalyzer, right? So it was like mixing together like a game with some functional utility. Just like I think we're
Michael (
40:18)
Yeah, that's cool.
Aaron (
40:19)
we're gonna see a lot of stuff like that. and I think some of it will be
Chris (
40:22)
Can only do
Chris (
40:22)
like boss battles when you're trashed, Aaron? What's the what's the mechanic here?
Aaron (
40:27)
I think it was just like, you know, somebody wanted to have some fun. So it was like picture like an anime breathalyzer. Yeah.
Michael (
40:35)
That's cool.
Michael (
40:36)
At Rising we do a lot of things with emulation, you know, like showing an old artwork. one thing we showed last year was Lori Anderson's CD ROM puppet motel. Actually she made it with a collaborator. and so to you know, you could like a live code a new version of the CD ROM that, you know, that looks good on a retina screen, but it kind of changes that sort of pretty thoroughly. So instead we we put it in
Michael (
40:59)
original software environment and then use an emulator to let people run it on a modern computer. But I'm just kind of like thinking that we need to like make some rhizome cyberdecks so we can sort of ship out like a
pri (
41:10)
That's smart.
Michael (
41:11)
a little hardware setup that runs these emulators really easily. yeah, I feel like that would be a fun way to kind of capitalize on this energy.
Chris (
41:21)
Yeah, but you still gotta give the kids a tactile sensation of of loading a mini disc into it.
Michael (
41:26)
That's true. Yeah, no, we need to
Chris (
41:26)
L little pop up little plastic pop
Chris (
41:29)
up thing, put it in, click. it spin.
Michael (
41:30)
Yeah.
Aaron (
41:30)
I mean,
Aaron (
41:31)
we may go back to that, Chris, because you know, my default's gonna be that anything connected to the internet's just gonna be under constant surveillance. So maybe maybe we do have to go old school to like zip drives and floppy disk drives on your on your open source hardware. Yeah,
Chris (
41:47)
All of a sudden Michael's life is gonna become about macromedia support.
Aaron (
41:52)
completely.
Michael (
41:53)
macromedia
Michael (
41:53)
is my favorite. But you like we're doing this archive space that with EAI at a new place in New York called Canyon next year. And it's we're thinking about like how to give people access to archival works in a physical space. And I was thinking about like an RFID system where you take like a VHS off the shelf, but it's actually just an RF ID tag. so you could do the same with a CD ROM, like.
Michael (
42:17)
Yeah, you you have to open the case and put it in, but then it's just it's not Yeah, exactly.
Chris (
42:21)
It's really just a key? That that's
Chris (
42:24)
I actually that I think that might have legs in a very strange sort of way. I mean it's it's it's it's full
pri (
42:30)
Wait, what how?
Chris (
42:31)
full like Badrillion, but well just I think like the return of physical ownership of media, right? Like that's a trend that just never clicked for me. You know, as someone who lived through all that growing up, like
Chris (
42:49)
There's no nostalgia for a cassette like having to fast forward or anything like that. But like the kids love physical media. you know. And so don't know. I feel like because maybe there's a break in in that whole lineage there, like you can just completely replace the functionality out and have physical media just serve a symbolic role.
Michael (
43:14)
But how good are those RFID tags anyway or whatever? Like peop they're really Yeah. Every time I go to an art installation that uses some sort of RFID it doesn't seem to work. Although I guess it works on the subway.
pri (
43:25)
I I kind of agree with you. You have to like yeah,
pri (
43:28)
you have like finagle it every time.
Chris (
43:30)
I mean, you can just etch the QR code onto the media itself in a way that, know, you're just scanning it for access.
Michael (
43:37)
Yeah, totally. That would be also an aesthetic crime though.
Chris (
43:41)
Against QR codes, or are you QR code purist or
Michael (
43:44)
I no,
Michael (
43:45)
against the against the physical media's nostalgic qualities. It'd be a crime against
Chris (
43:50)
Yeah, all right.
Michael (
43:51)
nostalgia.
Chris (
43:52)
I see you work for big big C D player.
Chris (
43:55)
You're just a stooge for Sony Corp, aren't you?
Michael (
43:58)
Macromedia, thank you very much.
Chris (
44:00)
Should should we put this week's episode on on Dreamweaver? Or should it be on Flash?
Michael (
44:04)
Yes, yes.
Aaron (
44:06)
I mean, you're
pri (
44:06)
I'm too young for these references.
Aaron (
44:07)
you're joking, Chris, but I a lot of people I mean, that's how I learned how program was Action Script, or started to learn how to program was Flash Action Script and learned HTML by building stuff in Dreamweaver and viewing the source code. So definitely was important.
Chris (
44:22)
Now in my household there were two paths to
Chris (
44:24)
technology. Like my wife was definitely very macromedia coded and I was like SQL Server. SQL Server boyfriend, Macromedia girlfriend.
Michael (
44:32)
How nice.
Michael (
44:35)
You went to different rooms at South by Southwest Interactive.
pri (
44:38)
Are you in a Marfo, Michael?
Michael (
44:40)
I'm going to Marva for one day.
pri (
44:42)
what? You're going Are you serious?
Chris (
44:43)
Marfa's really far away for one day. He got
Michael (
44:45)
I know. That's
Michael (
44:47)
how much but see, that's I have one day. So, you know, I went to Bangalore for one day once.
pri (
44:52)
Holy shit, really?
Michael (
44:54)
Some I mean, when Dario goes to a place he goes for one day, he's not hanging
pri (
44:58)
Yeah, but
Michael (
44:59)
out.
Michael (
45:00)
That's how it is.
Chris (
45:00)
It's Bangalore,
Chris (
45:01)
like one day in Bangalore is really like only two meetings with the traffic over there, dude.
pri (
45:05)
Yeah.
Michael (
45:06)
We did an event there. It was really cool.
pri (
45:09)
whoa, you did a rhizome of it in Pangloor?
Michael (
45:11)
We did, yeah. It was part of our when we did a partnership with Anthropic last spring and and they invited us to organize two events there.
pri (
45:20)
cool.
Michael (
45:20)
Yes.
Chris (
45:21)
I'd
Chris (
45:21)
love to get back to Bangalore. I like that place.
pri (
45:24)
Should we do a group trip to India?
Chris (
45:25)
Can we say it's a Leela?
pri (
45:27)
Yes. Also I
Aaron (
45:27)
I mean there's e
Aaron (
45:29)
there's Devcon in Mumbai in a couple of weeks.
pri (
45:32)
Devcon and
pri (
45:33)
Edge City in Goa in November.
Aaron (
45:35)
really?
pri (
45:36)
Yeah. I like yeah, I feel like if I had more flexibility I'd definitely wanna go.
Michael (
45:41)
Yeah, I really wish I could have stayed more than one day. No.
pri (
45:45)
I don't think you get much of a glimpse then, but that's cool. I didn't know you guys did that.
Michael (
45:49)
There was such a community there of like creative code people and like processing artists. We know a lot of artists actually in Bangalore. And now we have a city fellow working in Bangalore for us one day a week, which is cool.
pri (
46:04)
cool.
Michael (
46:04)
So definitely trying to build some connections.
pri (
46:07)
I you know, I happen to be Indian, so if you ever wanna chat, you know.
Michael (
46:10)
Huh. So I hear you're from India.
pri (
46:14)
I guess
Michael (
46:14)
What were we
Michael (
46:15)
just talking about?
Michael (
46:16)
Macromedia.
pri (
46:16)
old school old school tech.
Chris (
46:16)
physical media.
Aaron (
46:18)
do think old school tech is it's like everything old is new again, but you know, I think what we're gonna see to hardware is the same thing we s we saw with software, it's p proliferation, it's gonna get easier. I don't think it goes to a hundred, I think it could go to infinity 'cause that's kind of the last thing that we'll need to make 'cause software's effectively solved or pretty darn close to it. Like at least for what we can add to it.
Aaron (
46:43)
Yeah, so I don't know. I just think we're gonna see a lot more of that. I do think we're gonna see much more invasive government responses to AI. and probably the end point of that is just, you know, some form of know, surveillance of every connected device. which I think will make some of these like nineteen eighties hardware conversations, like kind of the cypherpunk type stuff, like a lot more relevant again. So that's what I see.
Michael (
47:07)
Yeah, totally.
Chris (
47:07)
You're going Deep
Chris (
47:08)
Max Headroom?
Aaron (
47:09)
A little bit, yeah. I mean I just think that I I think that that's the end point for like the AI safety debate. They're not gonna ban it 'cause it's just too hard, too hard to coordinate. but I I don't think people will care if the government's like listening on every device, just like they they didn't really care when that happened to the internet. advocates obviously might care. I may I'll probably care
pri (
47:25)
Yeah. Snowden Snowden outrage
pri (
47:29)
last it lasted like six months.
Aaron (
47:32)
Yeah, it's just like the it's just like probably the cleanest like solution to the extent that that that problem's accurate and not just like a power grab. and then, you know, I do think that means that there's gonna be kind of a not a retreat to touch grass, but like to just have non connected, very secure hardware. which I think then will get open source. So you'll see open source hardware. I think the same thing will happen as robots get better.
Aaron (
47:59)
People are not gonna want like their proprietary Tesla robot. They're gonna wanna control like the robot that they have, you know, give it more autonomy for itself. And so the wheels of time will continue to turn. And Chris, I'll meet you
Michael (
48:11)
Mm-hmm.
Aaron (
48:12)
at the anthropic oracle of Delphi, where we can sift through the entrails
Michael (
48:15)
Yeah.
Aaron (
48:17)
of some history problems or something like that.
Chris (
48:20)
I don't know man. I wanna go to Rhizome and meet you inside a Faraday cage showing an artwork where we can have a conversation that isn't monitored.
Aaron (
48:28)
Yeah, that I mean that that's probably yeah, maybe we should find the Faraday cage manufacturers and make some bets there.
Chris (
48:33)
You're bullish
Chris (
48:34)
bullish on Faraday cages?
Aaron (
48:36)
feels like you may need to be. So
Michael (
48:39)
Yeah, it's gonna I d we're we're entering the age of the intranet. I think that
Aaron (
48:43)
Yeah. I mean
Aaron (
48:44)
look, China built one of the best intranets ever. So might as well
Michael (
48:47)
It's really true,
Michael (
48:48)
yeah. But it
pri (
48:49)
True.
Michael (
48:49)
yeah, it does feel like AI is making this kind of geopolitical play where the AI industry is so is so big, it kind of it kind of can exist without government involvement on some level, you know? and people s there's a lot of I think so people will of course like be involved in and using and
Michael (
49:11)
educated through it and all kinds of complex new things. So I think a lot of the things we're describing are are a little bit of like a reaction against that too. Like people trying to claim autonomy
Aaron (
49:21)
Yeah, completely.
Michael (
49:22)
and trying you know, create the alternative space for something that
Aaron (
49:27)
Yeah, it's like that little
Aaron (
49:28)
movement that folks are starting to like put away your phones. I don't know if you guys have been following that. The Times were in like a article about that.
Chris (
49:35)
Isn't that like an evergreen story?
pri (
49:36)
I hate those people.
Aaron (
49:38)
It is, but I think it does feel like, you know, it is like picking up like a little bit more steam. Like it didn't feel like like a fake trend story which the Times specializes in, sometimes. But I don't know, it feels like a little bit more real to me.
Chris (
49:52)
I mean, I would say I barely ever use my phone, but that doesn't make me any better because I'm always on my laptop. You know? I
Michael (
49:58)
Yeah.
pri (
50:00)
I know.
Chris (
50:01)
I just need like a bigger inner screen to stare at all the time.
Michael (
50:04)
Yeah.
pri (
50:05)
My
pri (
50:06)
my friend was telling me in China that like screens are like super normalized so people just like have dinner and they have like a little spot for their phone that they could put in people like in the middle of conversation at dinner will just like pick up their phone like it's like super normalized and like not considered rude or offensive there. And I'm like, I love that. I love my phone. But
Chris (
50:23)
Yeah, well wait till you're a parent and you're taking your kids out to dinner.
pri (
50:26)
does that suck?
Aaron (
50:27)
Yeah.
Chris (
50:27)
I mean
Chris (
50:28)
it's just living in the twenty first century. But yes, it's annoying isn't
Aaron (
50:30)
Yeah, but
Aaron (
50:32)
it it's pretty much just like
pri (
50:32)
Okay.
Aaron (
50:33)
a constant battle to they're they're
pri (
50:35)
The tech.
Aaron (
50:36)
destructive and hype highly addictive devices. Like I don't
pri (
50:39)
Yeah.
Aaron (
50:40)
I you may love it, but you love it the same way you know people love to OD on sugar, you know, twenty years
pri (
50:46)
Damn.
Aaron (
50:46)
ago.
pri (
50:47)
You guys sound like my husband. It was just like you're heavily addicted to your phone. Like, please get a grip.
Michael (
50:52)
Yeah.
Chris (
50:54)
All right. Well, Free, we don't want to turn like the Society into a scolding of your phone usage. And
Michael (
50:58)
Yeah.
Chris (
50:59)
I'm certainly not about that.
Aaron (
51:00)
I mean you do you
Aaron (
51:01)
pre, yeah, but just know is you're gonna
Michael (
51:03)
Yeah.
Aaron (
51:04)
get like you know, iPhone diabetes or something like that.
pri (
51:08)
I mean I did have to get a little
Chris (
51:09)
iPhone diabetes. That's our episode title right there.
pri (
51:12)
I did actually have to get a little like brace because of my because of my like my phone like my fingers were so like tightly round like wound around my phone that like it was causing like joint issues for my fingers. So I had to like get a little brace.
Aaron (
51:27)
Yeah, I
Aaron (
51:28)
remember that. You tried to blame that on me. I was like, What are doing?
pri (
51:30)
It was partially
pri (
51:31)
because of work. Partially because of work. I should probably run.
Chris (
51:34)
Yeah, but now you're farming work out to instinct.
pri (
51:36)
I think that's also honestly part of it. Like I'm farming work out to Grokbutt that I'm like, wow, I'm like actually but now I'm on my phone on Grokbutt. No, I'm not.
Aaron (
51:42)
I think I think Pree's cooling on instinct. I think she's cooling on instinct.
pri (
51:46)
No, I'm not.
Aaron (
51:47)
You sure?
Chris (
51:48)
I
Chris (
51:48)
actually am loving Muse. Muse is like real deal.
pri (
51:52)
I need to do
Aaron (
51:52)
Yeah.
pri (
51:52)
musmore.
Aaron (
51:54)
Yeah.
Chris (
51:54)
It
Chris (
51:54)
is like it's it's kinda wild because I feel like from my like I'm testing all my APIs right now, which is just so tedious. I can just like have Muse like run through them all and then it prepares little reports for me. I'm like, Jesus Christ, the luck is like giving me Accenture here. Like what the hell is going on?
Aaron (
52:11)
Yeah, I saw
Aaron (
52:11)
you I saw you post that. That's crazy.
Chris (
52:14)
It was just like wild. Like it created this like whole perfectly formatted PDF and I was like, I asked you to test my APIs, man. But like great. Thank you. you know, Bain and Co. My my
Aaron (
52:23)
Yeah.
Chris (
52:24)
little guy who works for a a consultant.
Aaron (
52:27)
I think it Chris, like that that point I've been really I've been thinking about like what's the interface for agents and it I I increasingly think it it probably will be voice like you were describing, Michael, like a localized voice. You'll talk with it, it'll be like personified in some sort of way, whether we like it or not. and then they'll connect back to these like workhorse agents, whether that's like a Muse or a Crockbot. I think that's where instinct kind of gets squeezed out outside of the fact that they probably can't afford their infra.
Michael (
52:56)
It is kind of crazy how quickly that like that one way of interfacing with AI has become so dominant. The chat, you know.
Aaron (
53:05)
I think it's kinda like,
Aaron (
53:06)
you know, we're gonna discover the steering wheel for or like the search box for AI and it it feels like the form factors are gelling a little bit more. But I I don't think it's gonna be, you know, like a like a text interface that that feels too close to where we are now.
Michael (
53:23)
Yeah.
Chris (
53:23)
I'm I'm a little boomerish here and it's the same way like way back in the day when you were, you know, like at Trader Joe's and someone was just like randomly talking to themselves in an aisle and you didn't understand like they were on a phone. You know, like I'm a little that way still. I don't wanna be talking to my AI.
Aaron (
53:41)
So I think
Michael (
53:41)
Yeah.
Aaron (
53:41)
for your solo work you're not gonna want to, but if you think about most workplaces, you know, they're usually on calls, right? Like some for most businesses, most workers' days are like all on calls. And collaborative work is usually like a f if it's an in-person meeting, it's all voice-based. So I think you're gonna need the agents to operate really well in that. I don't think the end point is like you're sitting there as a team like, you know, messaging
Aaron (
54:10)
some sort of complicated agentic system through text, like through Slack or, you know, iMessage or something like that. That feels like too constrained.
Michael (
54:18)
Well guys I have to
Chris (
54:19)
Yeah.
Michael (
54:19)
admit I'm a little embarrassed here. I th
Aaron (
54:21)
Why?
Michael (
54:22)
I thought well, I thought we were all sending our agents to this call. I'm actually Michael's in the other room. He's he's doing something else
Aaron (
54:26)
probably could invite Aiden to do it to do the podcast.
Michael (
54:31)
right now. Yeah.
Aaron (
54:32)
Chris, should we invite Aiden to do a podcast? We can really mad max it up.
Chris (
54:36)
I think that's episode two hundred goals right there. It's just like we all send our our our AI representations to host a show. None
Michael (
54:44)
Mm.
Chris (
54:44)
of us get on. It's just Finkel
Aaron (
54:46)
Yeah.
Chris (
54:47)
and like four agents. No, we don't like Finkel
Aaron (
54:48)
F Finkel our agents
Michael (
54:49)
But Finkel doesn't know.
Chris (
54:50)
can send his agent too. Like we we can have like amazing production that show.
Aaron (
54:54)
Yeah, we'll be managing the you know the the sim the simulation that we all live in through our agents. So we can watch squirrels outside. All right, maybe on that note, because I know folk folks gotta run. Thanks, Michael, for for joining.
Michael (
55:03)
It was so nice talking to you guys today. I got some squirrels to check
Chris (
55:06)
Yeah. Thanks for coming on, Michael. Appreciate it.
Aaron (
55:08)
so that that's a wrap. you know, today we had Michael from Rhizomon, Chris, Derek and Pre. So welcome to Next Net Society.
Aaron (
55:19)
Just remember everything said here is not investment advice. It does not reflect the views of our respective employers. And thanks so much for listening.
Chris (
55:29)
Hold on, Aaron, I am gonna give some investment advice. Based on everything
Aaron (
55:32)
What's that?
Chris (
55:33)
we've talked about on the show, I think there's a gold mine to be had in buying old DVD
Aaron (
55:37)
In Faraday cages?
Chris (
55:39)
copies of Enemy of the State,
Aaron (
55:41)
Yeah.
Chris (
55:42)
in which there is a Faraday cage scene.
Aaron (
55:45)
Course there is. Of course. Alright, there you go.
Aaron (
55:48)
Outside of that that very narrow commodities recommendation. Alright.
Michael (
55:53)
Yeah.
Aaron (
55:54)
Bye guys.
