25:04
Good morning, John. So, there's this idea among people who study media called audience capture, which is when like a creator gets signals from their audience about what's most interesting to them, what angers them, what gets them most excited, what they like and what they don't like, and then they just like slowly become that perspective. More and more forever, the creator becomes a thing that isn't really them, maybe isn't even really coherent, and the audience and the creator reinforce each other's certainty. I've been searching a lot for certainty a lot, not just this last week, but like this last couple of years, it's felt destabilized in general. And so when I see dynamics like
that bring people to a certainty that I don't think is deserved by reality, you know, if you look at our previous ability to predict the future, it hasn't been great. So I worry about that. But I also don't think that creators have like zero obligation to their audience. Obviously, I hope the way community actually works is we all listen to each other. Not just to hear the takes at the surface, but the stuff deeper down. Not so that we can say, "Okay, Steve has identified the one true place that we all need to be." So, everybody get to where Steve is, but instead so that people can say, "I get where you are." And understanding that has moved me. Now, you're not going to get moved to a place where you have
exactly the same perspectives as anybody else, but we move each other. That is how this community has always worked. and it's how our brotherhood has always worked. I should say, John, you're a good brother. I hope that you could hear that. I appreciate you so much for being frustrated with me and worried about me and being kind with your frustration and making it clear that your compassion and your love for me is the biggest piece of whatever you're feeling. Also, I appreciate you not saying, "Oh my god, I cannot believe that this is distracting from the imminent release of the novel I have been working on for a decade." I appreciate you not saying it. I'm sure you are feeling it. So last week I
posted publicly about how I have been so enticed by LLMs. Always starting on a new project or prototyping a game or researching climate change regulations and 18th century organic chemistry and the history of screwdrivers all at the same time. The more time I spent with it, the more every problem was starting to look like an LLM shaped problem. And that's dumb. Of course, because of how online discourse works, some of the response has been silly or bad faith criticism. No one is using AI to write crash course videos. No matter what an AI generated headline said, Complexity's AI policy is years old. It mentions barred in it. So that's how old it is. And that policy is extremely clear. And
this is the thing that I feel most bad about. Whatever reputational hit or loss of trust I have brought up on myself here, that's mine to take and try to earn back. But the people who work at Complexly, who write Sihow and Endlings and Eons and Crash Course, and who work with experts, and make sure that they get everything right, do the video editing, make the visuals beautiful, do that hard work of putting together some of the most valuable videos in the world are now being called into question because of my issue. So, this video is two things. First, it is an apology to the people who work at Complexly. I'm so sorry. Like, you are so good and I'm so lucky I get to work with you and I see
how hard that work is. And of course, I'm really closely associated with Crash Course and Sihow, but there are whole teams of people who have each other's backs who are just trying their damnedest to make great things that help people understand the world. I get that not everybody's going to get this and I understand that what I do, no matter what context I do it in, is going to reflect on Complexity and so I should be more careful and I should not be on Twitter. Everybody agrees I should not be on Twitter. Second thing, following Complexity's lead, an AI policy. This all started because I said I needed to realign the way I use LLMs. I have thought a lot about that this week. I'd
become too reliant on them for advice, for editing, and for sourcing. So, I'm creating an AI policy for myself. One of the things I've been hearing in the discussion around this, and I'm glad this discussion is being had is that this is a good idea. I encourage creators to do this. It doesn't even have to be public. Just have it. I wish I had it. Another thing I've been hearing is that people don't want me to make stuff because like I'm rushing through a list of boxes I need to check off a list. I have never felt this way. Like sometimes I will want a video to be more elaborate than it ends up being because I need to be done with it. But I only make videos that I think are valuable. I don't like have a schedule
that needs to be kept. I stopped doing brand deals at the beginning of the year on Hank's channel specifically because I didn't like having that lock in. So it's not that I don't want to make stuff. I want to make too much stuff and I am convinced that it is possible. That's the problem. Whether it's an attempt at like legitimate analysis of the very weird world that we find ourselves in or it's just goofing off, I want to make that stuff. I like making videos, but I want to make sure that I like making them, not just having them be done. But that I have heard from people that they feel like I might just be checking boxes, that's a fault all in itself. And I think there's lots of opportunity for
me to be more choosy in a way that results in like only good outcomes for like me as a person, for the content, and for the audience. Like ultimately, let's be honest with ourselves, there's not too few YouTube videos. I still don't have the certainty that I would love to have about where we are right now or where we're heading. But having sat for a week and listened to this conversation, I feel more aware of what my job is right now. And a huge part of that is to be myself. Making more things does not make me make better things. And people need to be able to absolutely trust that my takes are my takes, which they are. And I know that there are a variety of perspectives on what AI
should and shouldn't do, but for me, you can see the policy in the description. Everybody, thank you. John, I'll see you on
