AI Detection Can't Measure Meaning: What It Actually Sees

Summary

Podcast Summary: "AI Detection Can't Measure Meaning: What It Actually Sees"

AI News & Strategy Daily with Nate B. Jones – July 22, 2026
Guests: Nate B. Jones (Host), Chris Best (Co-founder & CEO, Substack)

---

Episode Overview

This episode dives into the issues of AI-generated content, the challenges of distinguishing meaningful human creation from "AI slop," and Substack’s approach to preserving the quality and intent in digital public squares. Nate and Chris explore the blurry boundaries between responsible AI use and content that erodes value, examining both technological detection tools and the cultural norms they wish to foster. The conversation is candid, thought-provoking, and rooted in their direct experience as creators and platform stewards.

---

Key Discussion Points & Insights

1. Defining "AI Slop" and Its Threats

• Slop is defined as content nobody believes in: spam, clickbait, and "generic soulless stuff."
• Chris Best: “The thing that we exist to do in the world is help people make things they believe in and make money doing it.” [00:31]
• Both guests discuss a widespread sense of online content deterioration, leading to diminished trust and engagement.

2. The Rise of AI-generated Content

• Alarming statistic: According to Pangram, about 40% of longform writing on LinkedIn is AI-generated. [01:29]
• Chris emphasizes: “Not all AI is slop and not all slop is AI,” but the share of machine-generated text is rising in digital spaces.

3. Human Intent vs. System Gaming

• Intent is the key dividing line: using AI as a tool to enhance one’s message versus using it to flood platforms for cynical gains (e.g., SEO farming).
• Chris likens extreme mass AI content generation to a “denial of service attack against the public square.” [05:01]
• The challenge: Is this real? becomes the reader’s constant question, eroding trust and willingness to engage with new writers.

4. Substack’s Response – Transparency, Not Policing

• Substack integrates Pangram's AI detection, allowing readers and writers to scan for likely LLM-generated text. [07:56]
  • Not positioned as authoritarian, but as a transparency step.
  • Aims to help users make informed judgments rather than banning AI outright.
• Writers can optionally add a “How I make this” statement to describe their creative process. [10:03]
• Chris: “We’re not coming here and saying what you can and can't do... we’re just introducing one tick of transparency so people can see what’s happening.” [11:17]

5. The Limits of AI Detection and The Need for “Meaning Detection”

• Nate describes using LLMs extensively, but with rigorous intent and iteration, pushing AI to match his unambiguous vision or help clarify unclear ideas. [14:21]
  • Example: “You wrecked my clarity. You sawed it off.” [15:00]
• Chris observes: Even thoughtful, high-intent AI use can’t be robustly distinguished from low-effort mass generation by today’s tools.
• Both agree on the dream of an “idea graph”—a Pangram for ideas that maps originality, variation, and conceptual territory, rather than just word patterns. [22:35]

6. Value at the Edges & the Role of Community

• The most valuable human contribution is on the “edges of the distribution”—original, spiky, or challenging ideas, not AI’s central average. [24:28]
• True progress, connection, and cultural change come from these edges.
• Communities (digital and physical) are essential for nurturing conversation, surfacing new perspectives, and validating human creativity.

7. Rewarding Real Excellence & Anti-Slop

• Beyond slop detection: How can platforms like Substack highlight and reward exceptional, thought-provoking work?
• Nate suggests recognizing when posts "spark a ton of conversation" or deeply shift readers' perspectives, akin to the stickiness of Paul Graham or Naval’s essays. [31:25]
• Chris: “What is the opposite of slop, and how could we find and reward that?” [30:00]

8. Impact of Video & Evolving Norms

• As AI video generation improves, norms may need to adapt further. Video currently helps guarantee “investment of effort,” but this may not last. [39:06]
• Chris: “If I found out right now that during this whole conversation, you weren’t really here and I’d been talking to your avatar... I would be upset. It’s not because I’m anti–cool avatar. That would be a mismatch of my expectation.” [41:24]
• Evolving expectations will shape how authenticity and effort are recognized across media forms.

9. The Last Scarce Resource: Human Attention

• Chris: “What’s valuable is human attention... that’s not susceptible to magical 10xing every year.” [43:31]
• As distraction becomes abundant, allocating attention well—especially for children—becomes a core societal skill. [44:50]

---

Notable Quotes & Memorable Moments

• Chris Best: "The problem with mass AI content is, it's effectively like a denial of service attack against the public square." [05:01]
• Nate B. Jones: “If I have to push the AI this hard on ideas, really, implicitly, there's some sort of averaged, centralized distribution of ideas that AI tends to favor.” [20:01]
• Chris Best: “When I want Claude's opinion, I'll go talk to Claude. I don't need you to be a thin wrapper on Claude for me.” [23:13]
• Nate B. Jones: “The alpha in terms of ideas is absolutely at the edges of the distribution. I would also say that the alpha in human connection is going to be at the edges of that distribution.” [24:28]
• Chris Best: “What is the opposite of slop? And how could we find and reward that?” [30:00]
• Nate B. Jones: “Those kinds of viral ideas are not always viral in the sense that they make it mainstream and the algorithm always rewards them, but they're viral in the sense that when you find a reader for them, they kind of stay in your head all day…” [32:17]
• Chris Best: “The problem is not that you use the technology. The problem is that you betrayed the expectation… it's happening with writing, it's going to happen with video, we have a bunch of expectations we don't realize we have, because it's never come up before.” [41:41]
• Chris Best: “What’s valuable is human attention. And what’s non-inflationary is human attention.” [43:31]

---

Key Timestamps

• 00:31 – Defining "slop" and Substack’s mission
• 01:29 – Alarming rates of AI-generated longform content
• 05:01 – Mass AI content as denial of service on public dialogue
• 07:56 – Substack’s integration of Pangram, promoting transparency
• 14:21 – The drafting process and wrestling with LLMs
• 22:35 – Dreaming of a Pangram for ideas rather than just text
• 24:28 – Value at the “edges”: human originality and connection
• 30:00 – How to highlight and reward real discourse and “anti-slop”
• 39:06 – AI and video: authenticity, expectation, and new norms
• 43:31 – Human attention as the ultimate, non-automatable resource

---

Conclusion

This episode is a must-listen for creators, executives, and AI practitioners invested in the future of meaningful online dialogue. The conversation lays out both the urgency of confronting “AI slop” and the nuanced, evolving strategies for reclaiming human intention, creativity, and trust in the public square. Substack’s commitment, exemplified by the Pangram integration and spirited openness, is toward fostering a space where excellent, thoughtful work rises above the churn—ensuring that the final scarce resource, human attention, is honored and well spent.

Transcript

[A]

Chris, it's great to have you on here. Tell me about you, about yourself, how you got into this problem space. Would love to hear a little bit about it.

[B]

Hey, my name is Chris Best. I'm the co founder and CEO of Substack.

[A]

Amazing. And I know that we're talking a lot about AI slop today. We're talking a lot about what Substack is doing about that, how you're thinking about it. Maybe walk me into how you recognize this was something that you wanted Substack to address and sort of your perspective on the whole problem space and we can kick it around a little bit.

[B]

Yeah, you know, I guess in some ways Slop, defined broadly as, you know, content that nobody believes in is I kind of think of as the foil for Substack. The thing that we exist to do in the world is help people make things they believe in and make money doing it. And so, you know, defined broadly all the way from spam to clickbait to kind of like, you know, generic soulless stuff, that's sort of the stuff that Substack, you know, exists to provide a real alternative to. And then I've had the experience, I think many people have, you know, over the past 12 months on the Internet, having this feeling of more and more places that I spend my time and life online, feeling, feeling kind of like inundated with things that I perceive or I feel as slob. Things that are thoughtlessly made or that are presented as sort of deep work by somebody, but are actually kind of recognizably like Claude copy paste. So I sort of, as just like a consumer of the Internet, I kind of had this feeling and then there's been, you know, I started talking to the guys at Pangram who are one of the sort of leading, you know, AI text detection things, and they have, they had launched a Chrome extension where they could, you know, you can kind of like turn it on and see for yourself what you're reading online. And they put together some stats of kind of like, how prevalent is this on a bunch of major platforms? And the stat that kind of blew me away was for long form writing on LinkedIn, something like 40% of it was fully sort of generated according to their measures. And that's not a perfect, that's not like a perfect correlation. Doesn't mean that, you know, I think of it as not all AI is slop and not all slop is AI, but if you just think of, you know, the share of text that's human written changing dramatically in the Places where we spend a lot of our time online. It just seems very obvious to me that this is going to be a major. A major change and a major thing that platforms and people get to navigate.

[A]

Yeah, I very much think so. And I think that's why, like, when you sort of scouted the idea, the first thing I thought was, we got to talk. Like, we got to have a conversation about this. Because one of the things that's been, you know, really interesting for me is to see exactly the same trend line that you talked about, but to see it both from the wider Internet perspective and also from inside organizations. And so when I'm inside organizations now, you have kind of the same issue. You have docs that are written and no one knows what's in them. Sometimes it's slacks that are written, and you can just tell it's like the you're absolutely right fallacy. Right. You can just tell that there's sort of an AI smell about it.

[B]

Right. Somebody sent the PR to review, and you're like, I'm the first person that's read this.

[A]

I just know I am right. And I think that that's. It's a problem because, you know, substack, at least my perspective is substack exists because the public squares were thinking about the public square as something where we want to have a rich dialogue. We want to have human conversations. And that requires humans to show up and humans to think. And AI is certainly a part of that discourse. And I've certainly been at the forefront, sort of with my substack corner of talking through a lot of that. But if we don't bring our human intent into the picture, it's going to be really hard to convey thinking that has lasting value. And that's like, if I think about the foil that you're talking about, the contrast between slop and not slop, or slop and anti slop. It comes down to intent, and it comes down to thoughtfulness and engaging with the material. And then some people are going to use AI like a paintbrush and work within their intent and their thoughtfulness and try and create something. Some people are going to write it by hand. And either way, those can be different flavors or different places on the continuum. And there's going to be other people who are. I think the example you gave in our earlier conversation was really great. Like, they're going to tell Claude, make, you know, a thousand blog posts and make no mistakes, and they don't even know what's in them. And they're doing it because of, you know, an SEO play or whatever they're doing. And that doesn't contribute to the public square.

[B]

Yeah, that's the real dividing line to me. You know, I have no interest in policing what tools people use. You know, we're sort of anti, anti authoritarian, pro freedom, like give creators the means to do the thing. But I do think there's a dividing line between I'm using tools to make something I believe in versus I'm kind of gaming a system cynically to skip the step where I have to believe in anything at all. The extreme version is being, hey, Claude, write a thousand viral posts, make no mistakes, off you go. And the problem with that is it's effectively like a denial of service attack against the public square. Right. It's not just a problem of, oh, I made a thing and it was bad, even if it's good. The problem that happens is then as a reader, as somebody who's spending time online, you have to be constantly wondering, is this real? And under those conditions, it's hard to discover and find value in new things. Like, as I've been having these conversations, I've talked to people that have told me things like I stopped reading new writers or I had to stop going in my comment section at all because a few times it happened that I would get sucked into some long conversation with somebody and it slowly dawned, dawned on me that it wasn't a person at all. It was kind of somebody had done this thing that said, hey, you know, do some replies and max out my account. And I think under, you know, there are some people who are sort of anti AI overall and who think, you know, only humans should ever do anything. You know, we have space for those people, but we're not. I'm not coming at it from that perspective. You know, the way I see it is like, hey, AI gives us new superpowers, but it also gives us brand new ways to be annoying on the Internet. And we're just going to have to deal with that. Yeah.

[A]

So on that note, I think when you look at the problem of AI, it is notoriously sticky or difficult to separate out the use of the tool from the responsible use of the tool. And so I'd be curious how you're thinking about essentially nurturing the public square, which, to be very honest with you, part of why I came to Substack was I liked the quality of the discourse on Substack when I joined and I was like, I feel like this is a place where I could call home. I feel like this is a place where I could write and it would feel like it was the right square, the right discourse, the right channel, the right place for me to have a conversation at a thoughtful, strategic, high context level. And I want to keep it that way. And I know you do too. And so I'm curious, like, when you think about the sort of differentiating tool use from responsible tool use, how do you think about that?

[B]

Maybe the biggest thing is I do not feel like we have all the answers or even should have all the answers. This is kind of like a new. We're all kind of entering this new world together. And at the same time, you know, because things are moving so fast, I don't think we have the luxury of waiting around indefinitely and kind of like, we'll just see how this all shakes out. Because I think you could, you know, you could wake up and have the Substack app feel like LinkedIn very quickly. And so the thing that we're, you know, I think this will be out by the time this airs. The first step we've taken is to build in an integration into the Substack app with the Pangram tool. And so if you're somebody, if you're just using the app as you would, there's no change. But if you're somebody who's curious, you can go to a long form piece of text and basically do a scan that says, hey, according to the Pangram estimate, do you think this text was, you know, you had AI generated energy, things written by hand. And crucially, that's not the same thing as was this made with deep care or not? And it's not. Was this even AI used in various parts of this. The thing that the technology allows us to do with like a reasonably good accuracy is kind of just be like, you know, did this text go through an LLM as it was generated? And my view is, hey, we're going to kind of just start with like, put that on the table. Let people, you know, let people see that fact for themselves and have an opinion about what that means for them. You know, my experience using it has been, you know, I think there are some times when it's. Almost everybody thinks it's totally fine. It's like, hey, I'm looking up some information from ChatGPT. You know, almost everybody does that, Almost everybody finds that useful. And then there's kind of like the other end of the spectrum. I don't know if you, if you, if you saw this when the open claw thing was really hot, but there was a guy that was having his open claw bot respond to text messages from his wife.

[A]

I did see that. My wife and I had a conversation about that.

[B]

And I think, you know, that's the other end of the spectrum. I think, like, almost everybody sees that as maybe not being the coach you use. And then the thing is it like. I think so there's kind of this spectrum of. In what spaces online is the. Is the sort of the contra. The implicit contract between reader and writer, hey, I'm going to make something for you that's like, carefully crafted, and I'm going to use a bunch of tools. And then are there other spaces where, you know, in the comment section or in the. In your DMs or in these other places? Like, I think part of my theory here is that the places that are responsible AI use should. Should stand up to transparency. Right. You should be able to say, yeah, I'm using. I did. I use this is how I use this. And the reader should be able to see that and say, this is cool. We're also giving a little. A little option for you to add like a. How I make this statement. So if you want to provide extra context about, like, hey, here's my processor. Here's what you can expect from me. You know, we're kind of starting with just giving some transparency on both sides of the relationship, giving writers the tools to check these things for themselves. And I kind of see it mostly as a way to sort of open this conversation. We're not coming here and saying what you can and can't do. We're not coming here and shutting anything down or doing anything like that. We're kind of, as a first step, we're just introducing one like tick of transparency so people can see what's happening.

[A]

Yeah. And sort of it's a step into the problem space, I think. And I think it's an important one because you don't want to be in a position in a year where you see the number and it's 40% on substack. And that would be. That. That would be bad for everyone, I think. And I will say I think it's really interesting to talk about the Pangram concept because. And it may be worth sort of fleshing out, but Pangram concept partly works by essentially learning that LLMs have what we would call a particular peak around the way they use language. And for whatever reason, reinforcement learning, pre training, etcetera, they have centered around a particular way of expressing language. And Pangram finds a way to measure that. But that is also Somewhat steerable with individual models and individual skills. And I saw someone who released a fine tune of a model that was speaking specifically, fine tuned to be more human. And they were getting like very low Pangram scores, et cetera. And so I think Pangram feels like a way to essentially call out transparently that this is something that we need to talk about publicly. It literally surfaces it into the public square. And I think it gives creators like me an interesting opportunity to talk a little bit more about how we create stuff. Right. I have over 20 different skills that I mess with in Codex and also in Claude that are related to writing. And a lot of what I spend time doing, I use whisper flow in my creative process. And I spend a lot of time essentially arguing with the AI around the way particular ideas are expressed and how clearly they're expressed and how well it corresponds to my intent as the person who's putting the piece out there. And I think especially with sort of what I've put together, I'm always looking to sort of incrementally move the ball forward on a particular corner in AI with what I write. And so I want to make sure that particular concept for the day is extremely clear. And some days it's workman like prose, right? It's not particularly soaring. No one's going to say it's the Declaration of Independence, but it's clear. And I find that the classic LLM isms, the sort of the dramatic contrast, the LLM self talk, the way headlines tend to be over dramatized a lot, that tends to get in the way of communicating to my audience. And that's part of why I have a bunch of skills that are essentially designed to work on weeding that out. Right? Like how do you get it so that it's cleaner prose and it sounds more like frankly good technical English.

[B]

And how, how hard do you find you have to push it both on the pros and on the ideas themselves. This is so.

[A]

So this is really interesting. I'm glad you asked that question, because those are two different answers, I think, on the idea itself when I come to the piece. And I have initial clarity in myself on the thesis and what I want to communicate and how to say will often be like a 10, 15 minute unbroken whisper transcript that I give to start things off, because I have so much in my head and I just want to get it out. And with today's LLMs, like with 5.6 or with Fable 5, it's just not a problem. They don't have an issue with that length of context or anything like that. And so that gives the LLM a flavor of my voice and the way I approach it. And then at that point the clarity is already there from the beginning. And it's just about working with the LLM to make sure it doesn't wreck the clarity. And sometimes I have to be that blunt. Like I will see drafts that I get back in that situation where I've been clear at the beginning. And I will say, you wrecked my clarity. You sawed it off. Because LLMs tend to be cautious by nature. And I'm part of what I'm trying to do is move the ball forward. And so if they're cautious by nature, I'm going to get a watered down version of my thesis unless I'm deliberately pushing it to actually listen to what I said. And so in that process, my drafting tends to be no, listen to my original vision, respect it, you're extending it, you're putting it into good written prose, we're cleaning up some of the verbal tics, et cetera. But it's fundamentally got to be as bold as what I put on the screen. And that process, that's what sort of the pushing looks like, being faithful to the original vision. Whereas on the other side, there are days when I am wrestling with a concept and I know there's something there, but my initial transcript in is not clear. And I know it's not clear because I'm still trying to find the idea.

[B]

Right. People might in the past have said that writing is thinking.

[A]

Yes.

[B]

Like you haven't actually, you haven't finished thinking about this is maybe one way to say that.

[A]

And I'm very transparent with codex about that. So when I, when I talk I'm like, this is not a finished thesis. These are the three or four threads we could pull on. These are some things maybe I don't want to go toward. And I list them out and then I ask it to go back and forth with me. And so then it's more using codex as a thinking tool so that I. It's almost co writing with me and people think, oh, it's co writing in the sense that it's finishing your thinking, but it's actually more giving me like the wall to bounce the tennis ball off of so I can like wrestle with the idea and come back to something that I feel really good and have high conviction on. And so those drafting processes are longer and less certain. I have been through processes like that where it will be all day with Codex and 17 drafts and all AI is doing at that point is helping me draft faster, because normally I wouldn't get through 17 drafts in a day, but now I can, and that helps me to iterate toward clarity. And then I can start to really push on the. On the pressure pros in the styling. But I, you know, if you're doing 17 drafts, you're pushing it, right? Like, you're just saying, this is not good enough. This is not good enough. This is not good enough. And it goes higher than that sometimes because I just, I. I really have an itch or it feels wrong somehow when I'm not pushing the AI to line up with the vision that I have.

[B]

Yeah. This has been one of the interesting experiences I've had as I've talked, because I've talked to a whole range of people across the spectrum of AI use, all the way from, like, I would only ever type things with my human fingers all the way to, you know, people. To me, the thing you're describing is, you know, what I would think of as, like, sort of like, high perspective use of the tools. Right. It's like, hey, I'm actually trying to. I'm, I'm. I'm trying to get my perspective out into the world. I give. I care a lot about the outcome. I'm using this tool. And an interesting thing I found is that everybody across that spectrum has had, like, a really strong worry about the AI slop quotient. Because the. The problem is. The problem is there's a low effort version of that that can be hard to distinguish at first blush, but that can kind of, like, take over. The commons can take over, like, can make it harder for that stuff to, like, be differentiated and break through. And it's. This is the part something we talked about that I think might be interesting to get into here. I think there might be a sci fi technology that I wish would exist that doesn't exist yet. That would be very. Might be, like, the perfect thing. Because, you know, the part of the thing with the Pangram thing we're giving you, we're going to give you, like, the ability to put down some notes of your thing. Because the thing that we're going to find is if you're somebody who, you know, does 17 drafts and thinks about it and bounces it back and forth, you may end up with something that is still, like, the text of the final thing is still AI output, right? Like the text according to the pan grammar or whatever is generated by the thing. But it's the. And it used to be the case that basically when you read something on the Internet, if you read a long thing, like it's just, it's almost by itself a proof of work. Like somebody has put enough thought into this to at least write it out. Maybe terrible thought, but at least somebody spent this much attention. And now it's actually like, it's hard to differentiate that from the person who just says, hey, write me 10,000 plausible sounding AI takes. Make no mistakes. I'm curious like how you think about that and what you think. If you think there's like a technological, what the technological root thing that we actually want is there?

[A]

Yeah, that's a really good question. I think so. I think in some senses a lot of the AI discourse is focused on the physical typing of the human fingers. Like where do the words come from either when I type them or when the AI magically types them. And I think when I have spent almost every day for the last couple of years wrestling with AI trying to push ideas, the thing that leaps back to me is if I have to push the AI this hard on ideas, really implicitly, there's some sort of averaged centraled distribution of ideas that AI tends to favor. There's a gravitational sink, if you will, around a particular subset of concepts. I think one of my favorite examples actually is if you tell AI to make any kind of list or formal document about artificial intelligence, it will always, without exception, have something on ethics and governance, whether you asked for it or not. It's just baked in. And I found that to be an issue in some of my pieces because not every piece needs to have that right. And I write about AI a lot. And so there's something that I actually have to think about, think about how I handle and where I emphasize that and where I don't. And as a human, I have a much, it seems like a wider conceptual palette. And I think that what I would love to see is a tool, a graph, something that we can build as a community that allows us to map the wider distribution of human ideation, the way we explore concepts, the different angles or valences that we take on concepts that are as diverse as all of us are, and then within that to map the much tighter distribution of the sort of LLM averaged out approach to ideas and thinking. And so this is not necessarily a measure of words, it's a measure of ideas. And I think if we can get to a solution there, that becomes a much more useful and aligned approach to the larger goal of having a clean public square, because we want a public square that's diverse. We want to have many perspectives. We want to have thinking that's original and takes different angles. And so why not measure for that?

[B]

Right? So the dream might be kind of a pangram for ideas. Rather than looking at the text, it looks at kind of like the underlying concepts. I think that would be useful because, you know, one thing, it's a tempting thing is for people to be like, oh, there's this thing that's average to the AIs. It doesn't mean that that thing is bad. Right. There's a. You know, the way I sometimes put this is like, Claude is great. I love talking to Claude. I talked to Claude a lot. I get a lot of value from Claude. But like, when I want Claude's opinion, I'll go talk to Claude. I don't need you to be a thin wrapper on Claude for me.

[A]

Right.

[B]

And so there's.

[A]

We want humans to be thick rappers. We want humans to have really strong opinions and to drive what they want out of AI, not the other way around.

[B]

And that might get more true as the models get better, not less. Right? It's not the case. It's like, oh, well, if it's, if the prose was just a little better or if the ideas were better, the model can get arbitrarily better and more powerful. And what that means is in the places where the model is deeply valuable, it is going to be like, that's not what we're going to turn to humans for anymore. Right. Anything where it's like, I could just ask Claude, people are just going to ask Claude, and that's going to be great for a bunch of things. And then the place that's going to have continued value, the places that are not in the central distribution, a lot of which may just be, you know, maybe noise or maybe wrong. Right. If you, if you, when you expand that landscape of ideas into higher variants, a lot of what you'll get is stuff that is mistaken or stuff that is like, you know, whatever. But that's also where you'll start to get like the real data of new things. Like almost like, you know, if you think about where the alpha is, the alpha is going to be sort of definitionally in things that are not that you can't just easily get out of the thing.

[A]

I would go even farther than that. I would say the alpha in terms of ideas is absolutely at the edges of the distribution. I would also say that the alpha in human connection is going to be at the edges of that distribution. That when we are Making a public square. Like we're literally referencing the physical public square where we all used to meet and that we can still see in some of our old towns. And I think that one of the things that's going to be really important is to have communities, whether digital communities or physical IRL communities, or a mix of both that facilitate that kind of human to human interaction. Because there's some magical third thing that we all know that happens when we have a rich conversation and a third perspective emerges because two of us are chatting together. And I love that. And I think that's one of the most powerful things about the public square. And to your point about comments, we don't want to lose that.

[B]

Yeah, yeah. We're a social species. And even if you can have kind of like superhuman media in every part of latent space from the model, people are still going to care about discovering and caring about and paying attention to those things together. There's like a. There's another layer that happens that is not, you know, even if you had a super, like, you know, strong AI that could create any work of fiction or any piece of thing in like the most convincing, most artful way, people are still going to care about other people experiencing and reacting and having real lives around these things, I think.

[A]

And we care about the stories that we create together. Like, it's been really interesting for me. I watched the Odyssey over the weekend and I've seen this explosion of discourse on Twitter about storytelling and where we get our stories from, the origin of our stories and human connection over the centuries, etc. And it got me thinking, like, in a sense, what makes the Odyssey powerful is that after, you know, 3,000 years or whatever, we still have to have this debate. We still have to talk about this story. We still have to go back and forth and try and make sense of this text. And to me, that's the mark of a great piece of writing, is that it sticks with you and it makes you have these human conversations. And I think the pieces that I look back on substack that I'm most proud of are not the pieces where I could point to and say, oh, I was right here, or whatever. It's the pieces where it got a conversation going. Like when I started to talk about OpenEngine and that got the conversation going around memory and how memory should be yours and it shouldn't just belong to any given AI provider. That's something where I think there was a human moment that sparked there that was really powerful, and that was more than I Expected. And that's what you write for. Right. You can't predict that that will happen every time, every. But that those are the moments that you're looking to create that spark a discourse and a ripple that's much larger than an individual piece.

[B]

Very cool. Well, I should ask you this. You're. You're a user of Substack, both as a. As a writer and a creator, and I hope as a reader a bit. If you could have anything for your experience around this topic, like, what would you want? What would make this. What would be the best way to serve human creativity on a platform like ours?

[A]

You know, maybe we set aside the wave, the magic wand idea graph thing. I think that's really cool. Maybe we say for now we put it to the side. I think it would be really interesting to have some kind of signal from Substack when an idea, a concept, a theme that's been published or talked about on Substack is gaining traction really fast and sparking a ton of conversation and that it's relevant. And I think we see this a little bit with trending topics on X. And it can be gameable, right? You can absolutely start to game that. But I think with Substack, what's interesting is that you have long form writing that these ideas can take time to unfold. I read a really wonderful piece that still sticks with me over a year later around Jeevan's Paradox and how it relates to different technological revolutions from the. From the steam engine on down to trains on down to etc. Etc. Like, it went forward to the telephone, etc. And it just stuck with me because it had this like, longer perspective that I was able to really marinate with and I was able to think with. And if we could find a way to basically mark a particular series, a particular post, a particular topic and say in this area, in this domain, this is advancing the discourse. This is helping us all think a little bit differently, challenging our thinking, helping us expose our assumptions. That's sort of the opposite of Penguin, right? That's a way of calling out and saying there's some real excellence here that we would want to call out and say that this is. This is featured or this is highlighted to some extent. Because I think in a sense as a reader, that helps me feel like there's other eyes than mine that are looking through Substack for that kind of spikiness or variance on the edge of the distribution. That's really interesting, unusual and high quality.

[B]

Yeah, it's like, what is the Opposite of slop. And how could we find and reward that? Like, how could we reward the stuff that's real? Which is different, you know, because even AI aside, standard attention seeking social media algorithms will tend to kind of like sand off and homogenize things and find the things that kind of like, best push certain base buttons or whatever. And the thing you're saying here is kind of like, okay, what's the thing we actually want? What's the thing that's deeply valuable? And can we get a stronger and more reliable signal of that as a benefit to the reader directly, but also to the, also to the creator who wants to make those things and can feel like, hey, if I actually take the time to make something deeply thought out that kind of pushes the discourse forward, I'm not just going to get totally drowned out by 10,000 people writing clickbait stuff, whether they, whether they use ChatGPT for it or not, which in

[A]

turn increases confidence that this is a platform for that kind of work. And this is a platform where that kind of work is highlighted. And I think that produces a sort of positive feedback loop when you're looking for fresh voices and you want to find new perspectives, etc. I think that would be really interesting.

[B]

What do you think are some of the best signals of that?

[A]

So I love that question. I think that the best signal that I've seen is you're looking for an ability to develop a coherent thought that acts like a, a little bit like a virus, like it's a net new thought, it causes you to pay attention to it, you can't get it out of your head, it sparks new connections there and everywhere. And I actually think that one of the good examples of this outside of Substack is Paul Graham's blog. And he just writes these things that when he writes them, they sort of stick in your head. Naval does this too. And you just look at the world a little bit differently as a result.

[B]

And

[A]

I'm hungry for more of that, because what I'm hungry for is challenging my own thinking, challenging my own assumptions, not making the same mistake twice, making sure that I read the world as it actually is and not necessarily as I want it to be. And those kinds of viral ideas are not always viral in the sense that they make it mainstream and the algorithm always rewards them, but they're viral in the sense that when you find a reader for them, they kind of stay in your head all day and you kind of can't get it out because it's so densely expressing a Perspective that puts a fresh coat of paint, a fresh sensibility on something that you may have stared at a thousand times. Right. And I think about this because I think one of the moments I had for this was actually from the Anthropic Engineering blog when they wrote about JSpace a couple of weeks ago. And I feel like that blog post got read and went viral in ways that I don't think were super relevant because it got sucked into the consciousness conversation. But to me, the most interesting part of it was actually thinking about how brains work and how LLMs work as computational solutions to problem spaces. And essentially like very narrowly what they're suggesting is that it may be a natural, evolved response to difficult problem spaces to use a workspace approach where you evolve essentially higher order bits of knowledge and compute with those bits of knowledge rather than the sort of base tokens that you're working with. And that's also happens to be one of the sort of approaches to how people think we do consciousness. I didn't think about the last time I took my last breath. It just seems to be working well. And I found that really sticky, right? Like, because that helped me to think about LLMs as essentially computing reality in ways that was emergent. And so I've just been kind of chewing on that ever since. And so that kind of a thing, like it went viral, I think went viral in a weird way, but I think the value is still there.

[B]

Right. You're almost looking for an indicator of how much the reader has changed.

[A]

Yeah, like, like did, did this stick with me? Right. Do I find that this was impactful for.

[B]

It is also interesting too. You know, I think some people read the anti slop stance as an anti AI stance. But it is really interesting to me that you don't, you know, you don't see Sam Altman or Dario or the Anthropic Engineering blog posting, posting clod slop. You see writing, writing a lot of really thoughtful stuff. And I think that's like illustrative that there is like a. It's not, it's not a question of you don't have to be anti AI. In fact, I think anti slop is a pro AI stance understood correctly in my mind.

[A]

Especially if you think of AI as something that we want to integrate into flourishing human societies, then being anti slop is something that is profoundly pro AI. Yeah, I love that. So as, as we look at that you. So we look at this conversation a little bit, I think one of the questions that I'm left with is, do

[B]

you

[A]

Feel like when you look at the creator community, you look at Substack as a whole right now. What is your sense of how creators are going to respond either to this initial foray with Pangram or to the sort of larger conversation that you want to kick off here? Because this is very much. As much as it's a dialogue among all of us creators, it's also a dialogue with Substack. And I'm curious, sort of how you anticipate that. Rolling.

[B]

I guess my real hope is to help spark a conversation here and to help kind of bring the topic into the open and have this. We can sit here and think about this at Substack all day, but there's this amazing community of people who are thinking about these things a lot, and I kind of want to like, get the. That idea on it. You know, we have all perspectives on the platform all the way from, you know, Sam, Chris wrote a piece, the title which was, if you let AI do your writing, I will come to your house and kill you. Yeah, that's right. All the way to, you know, people like you and others who are kind of, you know, doing these, like, frankly, pushing the frontier of what's possible with these things. My hope is that we can start to have a sense of what the new norm should be and what are the things that are actually going to solve for the underlying question of value and how are we going to defend against the cases where low effort stuff threatens both human and thoughtful AI authorship? You know, what is. What is the version of that that can sort of durably not just survive, but like, reinforce and help the. Help the real stuff win? That feels like the end goal to me.

[A]

Yeah, it's almost as if what we're looking for is that the conversation itself will be a inoculation against slop to some extent. It will provoke a discourse that helps us to think across the wide variation of anti slop and to think frankly, across the, I would say, significantly narrower distribution of slop. Because slop itself, to me is a very boring concept. It's notoriously averaged out, which is why Pangram can catch it.

[B]

And I do think that. I do think the Pangram doesn't. You know, Pentagram is there for the text. But I do think the stance that a platform has does matter to some extent. Like, it's interesting that LinkedIn has this. LinkedIn also has a feature where you can have AI write your stuff for you in the editor. Like, there's kind of like, oh, don't, don't want to bother thinking, like, click this button, we'll do it for you. And I do think that there's kind of, you know, just setting a little bit of like, how are we thinking about this? Which what do, what are we actually standing for and caring about? And what is the intent of the platform on a. Trying to reward and trying to encourage. I do think it's not that that thing by itself solves everything, but I think it does contribute to the culture of a place in a way that can be kind of, you know, self reinforcing and self fulfilling.

[A]

You know, we haven't talked about video much in this conversation, but I think there's two directions I'd be curious for your take on. The first is I feel like the way Substack includes video helps as a guarantor of effort right now in the current state of AI. So when I produce a piece and I attach a video to the piece, that's me talking, right? Like I spent the time to sit there and talk and I, and I said all those words and I thought about all of those things. And when we look ahead, I wonder if we're going to be having a similar conversation about video in a year or so where video is going to be good enough that someone who's saying a thousand blog posts make no mistakes can do. You know, 500 videos make no mistakes and they'll be able to get them.

[B]

And if you look at stuff that I get forwarded by people of my parents generation, we're already there for some

[A]

fraction of the population we are, especially for short form video. We're there. And I look at that and I worry a little bit because I think that video has been so certainly as a creator, it's been a powerful way for me to connect with folks. I think this kind of a conversation is something that works really well on video. And sure, we can grab a, you know, slightly cleaned up transcript of this and put it into an article, but it sings when we're talking because you can see the hands waving, you can see the emphasis in the voice and our faces and all of that. And I, and I wonder if we're going to be at a point where we want to be emphasizing that sort of human dynamism in video too.

[B]

Yeah, I think it'll be very interesting and I think we'll, we'll, I think we'll need new norms around a bunch of this stuff. Right? We'll need norms where, you know, we never used to have a norm that says you don't have the AI text Your wife for you, because that was never a thing. And then the minute it becomes technologically possible, there's kind of this moment where it's like, oh, there's a norm for that. Like, you know, maybe it's like, whatever I can. It can argue with the customer service person, and that's great. But, like, you know, if I found out right now that during this whole conversation, you weren't really here and I'd been talking to your avatar the whole

[A]

time, would be upset.

[B]

I would be upset. And it's not because I'm anti. It's not because I'm anti cool avatar. I'm just like, that would be a. That would be a mismatch of my expectation.

[A]

It would feel like a betrayal. Right. Because you're trying to have a conversation with a person.

[B]

Yeah, I'm a traffic conversation with a person. And it would feel like, you know what I thought was like, a mutual investment of effort and energy. And these things turned out to be like, this is why I called. I sort of think of this as the writing version of this, as clod fishing by analogy to catfishing, where it's kind of like, the problem is not that you use the technology. The problem is that you betrayed the expectation. And I think the thing that we're kind of happening. It's happening with writing, it's going to happen with video, is we have a bunch of expectations. We have a bunch of sort of implicit philosophical beliefs that we don't realize we have because it's never come up before.

[A]

And all of a sudden it's going to challenge.

[B]

The norm is challenged. And then it's like, wait a minute. Can you use Claude to text me? Can you use Claude to write a comment? Can you use Claude to have a video conversation with me? Like, when is it okay? Do you have to tell me? Like, these are just. I think there's gonna. They're gonna have to be, like, cultural outputs and expectations to these things through a process that I think is quite hard to predict. And I don't think the answer is going to be nobody ever uses AI to make video. I think it's going to be an incredibly powerful tool in all kinds of ways. And the thing we're going to have to find is kind of like, what are the ways that remain valuable? People talk about Jevons paradox a lot, but I also think a lot about Bommel's cost effect. You know this thing.

[A]

Yes, I do.

[B]

Where it's sort of like, okay, as some things get much cheaper, what remains valuable? Like what's the equivalent of a string quartet? And how do we guarantee that the like, how do we. What are the norms that support those things existing and being findable and being supported so that you know that it's good to invest your time in them because you're getting some version of the real thing. I don't know the answer for video, but that feels to me like the right question.

[A]

At least it does. And I think it gets back to the heart of what we've been talking about. About which is the idea that what's valuable in a world where to Carpathy's point, 99%, 99.9% of attention is AI. What's valuable is human attention. And what's non inflationary is human attention. What is not Susceptible to magical 10xing every year is human attention. And so how do we make sure that when the human has eyeballs on something, it is worth the human paying attention?

[B]

It's the last scarce resource. Yeah, there's a small. This has kind of always been the theory of substack. And we thought about this. You know, one of my. The story I tell myself about the early Internet and social media is that there used to, like, you used to be able to get bored. Yeah, there used to be, like when I was a kid, you could have this problem of like, man, I have two minutes right now and there's nothing to do. This sucks. And we've kind of like, even before AI, we have solved that problem. Any time you spend, you know, and before it's like, oh, give me something free to kind of like slightly distract me, it's like, oh, that's a great deal. Because like, I was bored and now I have something and now we live in a world where every piece of my attention that I could possibly want to spend, I can spend. And the question is like, how do I spend it well? And that's a very deep question,

[A]

but it's a worthwhile question because I think as a parent, that's what I think about for my kids is I'm like, am I teaching them how to spend their attention well? Am I teaching them the habits? Am I modeling the habits of how to spend your attention well that they can learn from and understand essentially what it means to be human by learning how we allocate attention toward one another in positive and constructive ways?

[B]

100%. I think you'll see more in person. What's that?

[A]

I was going to say Any final words?

[B]

Any final words? Subscribe to my Substack Check out the new feature I always tell everyone else to plug their stuff. I'm cb.substack.com I tell everyone to plug it, plug theirs. So I got to live up to that.

[A]

You got to live up to it. I love that.

[B]

I love that.

[A]

It's really been a pleasure, Chris, and I think it's a really important conversation. And so my hope for this video is that this becomes something where folks who watch this video will put it on YouTube, we'll put it on substack, of course, that we find a community of conversation around this topic. What it means to be anti slop in the age of AI and all the different ways that we can be anti slop. What it means to take a stance in favor of the human discourse in a world where slop is very prevalent. So I'm excited.

[B]

Fantastic. Thank you very much. This has been great.