Theo Was Called Out

Let's go for some drama. Command AI. WebDev Simplified released this video where he calls out Theo specifically, I mean brain rot with a picture of Theo there. He basically um calls out Theo in one of his latest streams or videos uh where Theo is quite brazingly to be honest saying you're using AI wrong. This is how you use AI. There's a bunch of concerns here. There's three core ones and they're all big and very detailed. But there is just so much info here that I don't feel like parsing. The same kind of thing he had with the other issue. He's talking about how there's too much for him to actually read here. So, he doesn't want to actually read and parse through that information to figure out what's going on. Now, I would argue if we just kind of go back in the video a little bit, you can see here it's not that much. It's maybe like one or two pages of actual reading that needs to be done for these different points. So, it's really not as much as he's making it out to be. But either way, he's saying, "Hey, this is too much for me to be able to figure out." I'm going to be pushing this off. And instead of just having the AI summarize it for him, and then him reading that summary, instead, what he decides to do is tell the AI to create a comment to post on that PR for him. Listen to what he says. This is fantastic research you've done. I want to leave a useful comment for the user who sorry for the contributor who put up this poll request. Help me draft a good comment to leave for this user with the core insights shared here to make it easier for this contributor to fix. So essentially, as you can see, he's prompting the AI saying, "Hey, good job doing this research. Can you now take this research, draft up a PR comment, and then later in the video once that PR comments gets drafted, he makes one revision to that comment, but it automatically gets posted to that PR. So, what he has done is he's taken someone's issue, he ran it through an AI to generate a response, and then he gave that response to that person as the output. This is a terrible use of AI. Again, you're constantly kicking the ball down the road. With email, people would send out an email saying, "Hey, can you handle this? Hey, can you handle this?" So would it constantly get kicked down until the last person available to finally do it. This is the same thing with AI. You're essentially taking this issue someone has, feeding it to the AI, getting some other information, and then feeding it back to that person. You're constantly kicking that ball down the road. Instead of actually just implementing or dealing with that feature yourself, you're just prompting an AI, telling it the information, and then the AI is doing that for them. This is work that does not even need Theo at all. Anybody in the entire world could take an issue, copy paste it into an actual input field for an AI, get the output, and maybe make one revision. He's talking a lot about this book, A World Without Email. The idea here is that with this email um with what like the analogy he uses um you you don't make any decisions. You you you you're supposed to have a meeting and you write an email and you say, "When do you have time?" And nobody like gives suggestion, nobody is doing anything productive. You are just um shifting away the work from yourself. You're pushing it to somebody else to another level and that is creating this chain of unproductive. I I guess this is the OG meat proxy. You you just shift it to send it to the other person. Oh yes, [laughter] I like this car. I swear I was I didn't read that. Yeah, I'm sure this video will result in totally rational and chill reactions on Twitter. So, with that being said, this is Theo's response to the to the video. Yeah, I think there's some like there's some um reasoning here and some justification here, but we'll get into it. Another day of unemployed Kyle aka WebDev Simplified dunking on me because he has no real career experience. His main complaint here is that I show multiple instances where I didn't know the depth of how part of the code bases worked. Sounds bad until you realize it's a codebase with over 300 contributors, millions of lines of code. Doesn't matter how much AI is used, nobody knows the implementation details of all the code code in a real world codebase like this. The example we highlight is one where I was unfamiliar with the details of how tagged skills are attached when we pass prompts to Claude. So, I'm guessing this was a T3 code um review or feature and they've got to interpret your use of skills inside of T3 code and say, "Well, are you using codeex or are you using Claude and send that skill to them uh in the in the format that they expect." So, there's a bit of a tree branching thing. Funny enough, I know a lot, probably too much, about how we tag skills and how messages are passed over SDK. But this one intricacy was new to me. Knowledge gaps like this are the norm in real work. Learning learning to operate with partial understanding of the system is key to long-term success as an engineer, regardless of AI usage. I'm genuinely scared that inexperienced devs like KL are going to trick well-meaning juniors into holding back their careers. In the video, Theo did not understand the or or chose not to try and understand, which he admits to hear that this was an intricacy that was new to him. Yet, he still kind of he he asked AI to summarize it and be a meat proxy in reviewing because it was a code review, reviewing the code and and sending that review back to the original developer who was developing that feature. So, he was a meat proxy in in that instance. And I'm sort of in between cuz I'm like on the one hand like we should absolutely be uh relying on AI to better ourselves, to learn and and it's going to get better. As we said earlier, you're going to get people who don't know how to do stuff and AI is just going to take care of absolutely everything. Right now, we're it's still kind of necessary to know how systems are put together and and various things like that. On the other hand, you know, there's lots of juniors, there's lots of um people learning coming to Theo's channel and saying, "Well, how do I learn how to do dev? How do I learn how to do use AI?" And and Theo's there promoting this idea of, "Don't worry about it. Run it through AI. Let it do all the problem solving for you." And it's like, you're not you're not going to learn like that. I I feel like it was a little bit of like uh the contrast between you and me. We had a series of videos where you were teaching me how to do a dev with AI, but not just just to be clear for the audience, not to Vive code, but actually do dev like from a dev perspective. And the main issue was always me not reading the plan or not reading it thoroughly and not understanding what I was reading and saying like cool go. uh which is always the case, right? If I'm building something where I'm out of my depth, uh obviously I can either sit through it and try to understand it or give the go. In my case, I I feel like uh a lot of times is I try to get an understanding but just broadly and it's not always the case. I have to admit it's not always the case. uh if I build like a landing page if it's anything like front end or like websites I I do understand quite enough for that but when it's when it's a bit more complex I I might actually let AI to do the decision making and I think it will come back at some point as you said like if you just say yes yes yes yes at some point you made some decisions and you have to face them. And these decisions might be structural. It might it might be something that hits you with a bill with a 2K bill actually. Like these things can happen if you um be become a complete meat proxy. So I get the point of like I get the idea why you should understand. Isn't it normal to be out of your depth and actually try things where you don't understand them? you still can copy stuff or like put things together without fully understanding how they work. Well, exactly. And I think that there in lies the problem. So, if we just chuck down here, you got Brad from Traversy basically saying the same thing um that I am in that ultimately the message here. I mean, Theo is a little bit butt hurt cuz he's being targeted and that's quite unfortunate. I I I feel I feel bad for him in that respect. But he is putting himself out there as a leader and then demonstrating a very poor form by not doing what you just said, which is this skill of reading and trying to understand. Theo didn't even try to understand one of them. He was like, you know, there's there's too much text here. And actually, uh, to Kyle's argument, like there wasn't a lot of text there. It would have taken him 30 seconds to read it. Interesting. If he didn't understand it, fair enough. Like, yeah, I agree with you. There's not you're not going to understand every little thing, but at least attempt to do it. Don't just meat proxy yourself and just be that's too much for me. Send it to summarize this and send it, you know, send it to the to the person. So the focus here really and the focus of Carl's video although it comes across as quite harsh is let's let's let's still try and understand because this is a that can only be a net positive on the world that you at least try to read and understand that you're not just this slave to yes accept except accept try to be better and and you know we we had the chat before the show about what what and maybe we can talk about this now, but what do where's the missing gap of people's understanding and you know they're see they're seeing a lot of people build applications they're seeing a lot of people launch websites and thinking wow they've done that with AI how the hell can they do that and it's because we've learned it's we're building on top of separate individualized learnings over time that allow us to be able to launch an app or a website. It's not just something that people have come to just it's a compound knowledge. It's compound knowledge. It's not that's very rarely does someone with zero knowledge hop into claude code or hop into cursor and build themselves an app right that they pro that most of these people probably should use something like lovable something where there's an infrastructure around that where you don't have to think about the little intricacies like database or authentication or hosting as simple as hosting or CDNs you don't need to worry about all that but By launching an app or a website within a a system like that, you'll start to pick up on things. You might hear the word CDN, as you say. You might read a response that says about a CDN, but all that's done is plant a seed to when you start seeing it over and over and over again. You might be inquisitive to look into what a CDN is, as an example. So, [snorts] it all stems from like the willingness to learn and and grow. And and I feel like Theo's video was wasn't exactly promoting that, which there's nothing like why would you not promote someone to try and try and learn, at least attempt to to read the response, you know? No, I get it. I get it. Uh Theo's respond also was a bit around like real world world knowledge that I want to also mention. like in a realw world scenario you might build something for someone or for yourself and not understand everything at that time. But I fully agree and I think it is good to even if you say yes to Claude um five times in a row, maybe you come back at the final once you have the final solution and you'd be like I actually do this. uh sometimes when when it's not too difficult for me to understand like how did you do it actually I do this as a like I want to understand what happened um and not just like partially because if you get you want to try something design something get somewhere and it takes you some tries it takes you some hours of work and if you learn how you did how you got there it will help you to get there next time much faster because now you know the at least at the very least the terminology you at least understand the dare I say the architecture to some degree yeah I agree it's good to learn yeah and and I was in agreement to be honest it it could it could have just as well been said without attacking another creator you know he was using as a launch mechanism That's that's serious. Yes. From YouTuber to YouTube. I mean, we we know that how it works. Like I don't love saying Web Flow is dead every week, but kind of like we're not attacking other creators though. We're we're talking about another business, right? We're not saying uh you know, take Timothy Rick as an example. We're not saying Timothy Ricks is stupid or you know, he was wrong or anything like that. We Yeah. True. Yeah. True. It it would it would be like me looking at a video from Timothy Rick and like because he makes a mistake or something and then I'm being mean to him. That that would be quite me. I mean uh we know in our community to do better. So maybe maybe the broader dev community can learn a little bit um from the Webflow community. This is part of a larger conversation on my show Command AI which we stream live every single week. We discuss the news and all things related to AI in the world of design and web. Catch us next week and join in the banter. Command AI.