206: The Vibe Coding Trap and What to Do Instead

Welcome to No Hacks, the podcast that cuts through the noise to reveal the truth about the future of work and the impact of AI. In this episode, we're taking a look into the phenomenon of "vibe coding" – the idea that you can simply describe an app to an AI and have it magically built.

Is Vibe Coding the shortcut to tech success, or just another false promise?

Sani argues that the hype around pure vibe coding mirrors the deceptive playbook of dropshipping gurus: selling a dream that ultimately profits the platform, not the aspiring creator. We break down the seductive promises, expose the harsh realities, and reveal the catastrophic failures that occur when the "vibes turn bad."

What you'll learn in this episode:

• What Vibe Coding Really Is: Understand the difference between responsible AI-assisted development and "pure" vibe coding, where code is accepted without full understanding.
• The Anatomy of a Hype Cycle: Discover the striking parallels between the vibe coding phenomenon and the dropshipping course industry, from their sales pitches to their hidden realities and who truly profits.
• A Catalogue of Catastrophes: Hear real-world horror stories of instantly hacked startups, data deletion disasters, and AI models that "lie"—illustrating the dangers of relying on AI without deep technical oversight.
• The 80/20 Trap: Explore why AI can get you 80% of the way to a prototype, but that crucial final 20%—security, scalability, and integration—requires uniquely human skills.
• The "No Hacks" Skills for the AI Era: We conclude by revealing the four critical, future-proof skills that will define the next generation of builders and leaders in technology: Systems Thinking, Problem Decomposition, Architectural Integrity & Security, and Expert Curation.

Don't fall for the illusion of "irrational confidence" in tech. Tune in to understand why deep, durable skills, not magic, are the real path to success in the age of AI.

Important links from the episode:

• https://www.reddit.com/r/cscareerquestions/comments/1jmyk5k/seems_like_the_guy_who_invented_the_vibe_coding/
• https://www.finalroundai.com/blog/vibe-coding-failures-that-prove-ai-cant-replace-developers-yet
• https://www.louisbouchard.ai/genai-coding-risks/
• https://medium.com/@lars_13145/system-thinking-and-ai-redefining-software-product-development-a193a08119bc
• https://instil.co/blog/critical-thinking-in-the-age-of-ai-and-why-it-still-matters/
• https://en.wikipedia.org/wiki/Lance_Stephenson
• Comparison table

No Hacks is a podcast about web performance, technical SEO, and the agentic web. Hosted by Slobodan "Sani" Manic.

Transcript

207: Vibe Coding
===

[00:00:00] Sani: Welcome to No Hacks, the podcast about the future work, the skills that matter, and the shifts AI is causing under our feet. I'm your host, sunny and today. We're tackling the elephant in the room, the phenomenon that is swept through the tech world, with the force of your typical Silicon Valley hype cycle.

[00:00:26] Sani: /

[00:00:26] Sani: It's a promise so seductive, it's going to feel like magic. So let me, let me paint a picture for you. Imagine you have an idea for a nap. You've had it for a long time, you're just not technically enough to build it. So instead of. Paying for the expensive evil developer or spending months to learn a programming language.

[00:00:45] Sani: You just describe that to an ai. You talk to an ai, write a prompt, tell it your vision. And like some digital genie, it gives you the software you dreamed of. No, that's not science fiction. This is the promise of vibe coding. But here a no hacks. I like to look at the truths behind the trends, so I'm going to argue that the hype around vibe coding is a mirage.

[00:01:13] Sani: It follows the exact same playbook as the drop shipping gurus did back in the day. Maybe still, I dunno, selling a dream that profits the platform and not the user. So it's vibe, coding. Just a drop shipping course for people too smart to fall for a drop shipping course. Let's break down the hype. Let's look at some of the catastrophic failures, and at the end of the episode, the four critical future-proof skills that actually matter in the age of ai.

[00:01:46] Sani: The ones that will separate the real builders from the Dreamers, the vibes, I don't care. So let's start with a clear definition of vibe coding. What is vibe coding? Really? Vibe coating is a practice of using natural language prompts to have an AI generate, refine, and even debug an entire application. The workflow is a simple conversational loop.

[00:02:11] Sani: You describe a goal, the AI generates the code, you run it. If it breaks, you just tell AI what went wrong and it fixes it. And you might even do that without typing, with using your voice. So you never read, you never write a single line of code. You never even have to write to prompt with voice mode. It is crucial to distinguish between two versions of this for this episode, and in general, it is crucial to understand that there are two approaches to AI assisted coding.

[00:02:43] Sani: On one hand, you have the responsible AI assisted developer. This is where a professional software engineer or someone with experience uses an AI as a powerful assistant, a pair programmer to speed up their work. They guide the ai, but they review, they test, they understand every single line of coded generates the the entire system, and they take the full ownership.

[00:03:08] Sani: You can assume, I'm not talking about this today. What I'm talking about today is. Let's call it pure vibe coding, so nothing but vibes. The defining characteristic of this pure vibe coding is that the user accepts the code that a machine writes, that the Vibe coding tool writes by asking a foundational model by.

[00:03:29] Sani: Open AI and TRO or Google to write it without fully understanding the code or understanding it at all. It prioritizes iterative experimentation over code, correctness or structure, and that's the version that's being sold to the masses. The one that promises anyone can build anything, but before you try it or go ahead and try it, but before you actually rely on it.

[00:03:54] Sani: Remember that every foundational LM that diesels use. Underneath every chat it says they can make mistakes. Keep that in mind, and that's where the danger lies. The very name vibe Coding is a masterful piece of psychological framing. It's a deliberate rebranding of a complex discipline of engineering into something that sounds casual, creative, accessible, fun.

[00:04:22] Sani: It semantically distances you from the rigorous, logical, and often frustrating reality of software development. Phrases like, forget the code even exists, and it's not really coding. I just see stuff, say stuff, run stuff, and AI does it for me. They're designed to lower the barrier to entry and expand the market to anyone who feels intimidated by the word programming.

[00:04:44] Sani: Now, this rebranding is made even more powerful by its origin. The term did not come from a marketing department. No, there, there are a lot of marketing departments in this software category burning a lot of money, but it did not come from them. The term came from Andre Carti, uh, one of the founders of OpenAI. And this. This gives it an immediate air of technical legitimacy. Promoters of vibe coding tools will constantly use his name to validate their products, and conveniently gloss over the fact that Carpathia himself frame his experiments as being for throwaway weakened projects.

[00:05:21] Sani: I'll say that two more times. Throw away weekend projects, throw away weekend projects. If you say it three times, a vibe coded production ready app might appear. It hasn't happened yet. There's a classic appeal to authority and it masks the enormous gap between the researcher's playful experiment because that's what it was, and a viable strategy for building a commercial product.

[00:05:46] Sani: And the parallels between the vibe coding hype, and the drop shipping course industry, let's call it that they're impossible to ignore. They follow the same script, they target the same human desire for a shortcut to success. So let's start with a sales pitch. A drop shipping guru will promise life freedom with minimal risk.

[00:06:12] Sani: Minimal effort. They'll sell a dream of low startup costs, passive income, a proven system that anyone can follow to achieve financial independence. Anyone can follow this to achieve financial independence. There are marketing. It's a highlight reel of supposed successes, flashy income, screenshots, rented Lamborghinis, luxurious villas in Bali or Dubai.

[00:06:36] Sani: All designed to sell you a lifestyle, something you want to have but you don't have it. Now, let's talk about the promise of Vibe platform. Any of them, they promise you you can build a. Minimum viable product in hours, not months. And you don't need any skills to do that. And they tell you to focus on your grand vision, not on the boring syntax.

[00:07:03] Sani: They offer. They say they offer one click deployment. They free you from any complexities, things you don't understand. You know, I, I, this is just too complicated for me. That kind of stuff. They sell you the dream that you don't have that problem anymore. They're selling you a dream of the empowered non-technical founder being finally unshackled from need to hire expensive evil engineers.

[00:07:30] Sani: Both of these pitches are selling an abstraction layer. That hides complexity. The drop shipping guru says, you don't need to understand manufacture, logistics, supply chain, anything. The Vibe coding platform says you don't need to understand computer science, security systems architecture. Both are selling you a shortcut that doesn't exist.

[00:07:53] Sani: And of course. It's a sales pitch in both cases. Always leaves out the inconvenient truths. The reality of dropshipping is brutal. The success rate is very, very low. The real work is not in setting up a Shopify store, installing apps and clicking a few times to create your funnel tracking with whatever Shopify app you choose, or they choose to sell you with their referral code.

[00:08:19] Sani: It's navigating the competition, finding reliable suppliers and mastering the incredibly difficult skills of marketing, branding, and customer service. Who's gonna run the ads? How do you know how to set up the ads in this scenario? How do you know they work? All for razor thin profit margins. Who's going to do the CRO?

[00:08:41] Sani: Who's going to make sure you, your website is not terribly flow? Good luck. Now, the hidden reality of vibe coding is just as harsh and who better to explain it than the man who came up with a term on or path in a later post? He compared his frustrating experience of trying to build a real world web application this way as.

[00:09:03] Sani: Being a bit like assembly IKEA furniture, an overwhelming process of piecing together and configuring a dozen different services, hosting databases, authentication, payment storage, and more. His observation is an indictment of the vibe coding dream. The second use trace just slightly from the getting started tutorial.

[00:09:25] Sani: From the documentation you end up in the wilderness. Getting started tutorial good. Anything else? Anything more complex, bad? And this is the critical truth that the vibe coding tools will never be upfront about. They operate on a predefined, happy path scenario. They can build the simple demo app, they can build something that looks like a template, but the moment you need anything custom, anything unique to you and your business needs, or have to integrate a new service.

[00:09:56] Sani: If you don't know how it works in the backend, you're lost. This is not a bug. This is just how it work. It's a feature of their business. These tools, they can't handle the real work of software engineering, the configuration, the plumbing, orchestration, workflows, best practices that separate a toy project from your real product.

[00:10:17] Sani: And that brings me to the core of this analogy in the world of drop shipping. The most reliable and profitable business is not running a drop shipping store. It's selling a course that teaches other people how to do it. Gurus can make or claim to make hundreds of thousands of dollars a month selling a digital product, a course to teach you how to drop ship, and that's far more stable and lucrative than dealing with angry customers or unreliable suppliers.

[00:10:46] Sani: The vibe coding ecosystem is very similar to this. There's an AI gold rush going on. Everyone wants to do something with ai, and there are very few venture capital backed startups. I'm not even going to say their names on this podcast. They're all competing for market share. This movement is probably not organic.

[00:11:05] Sani: It's probably a calculated market creation play by investors. The strategy is. To fund a handful, a dozen of competitors, fuel a media hype cycle, establish a new profitable software category. The business model is to sell you the aspiring creator, the non-technical founder, whatever you want to call yourself, a monthly subscription to access their AI tools.

[00:11:28] Sani: The primary beneficiary, the platform owners, their investors who are looking for a high valuation exit, not the end users trying to build a sustainable business. On their technology. 'cause let's say it again. Their technology is for throwaway projects. Disposable prototypes. They're both the dropshipping course guy and, and the vibe coding platform.

[00:11:51] Sani: They're the modern day prospectors, selling picks and shovels to the gold miners. So I have put together, and I'll share this in the show notes. I put together a table comparing different features of a dropshipping guru selling it a course and a vibe coding platform, selling your dream. The promise that the dropshipping guru gives you is you can become a millionaire overnight.

[00:12:16] Sani: It's easy passive income, the vibe coding platform. Build any app with no code, just focus on whatever else. The target, dropshipping, guru, it's aspiring. Entrepreneur with no startup capital seeking a shortcut. And for the vibe coding platforms, it's a non-technical founder hobbyist seeking a shortcut.

[00:12:39] Sani: They have something in common. The human reality is the huge failure rate for the dropshipping industry, isn't it? Industry. It should we even call it industry? I dunno. Logistical, nightmares, intense competition. That's not what they say when they're selling you the course. And, uh, for the Vibe coding platform, the reality, the hidden reality is unmaintainable code, massive security holes, architectural chaos, and so on.

[00:13:05] Sani: The real product for Dropshipping Guru is a course recycling free information available online and for the Vibe coding platform, a monthly fee to access a foundational model using API. That's all. So who wins? Of course, the guru selling the course, if they can sell it and the VC backed platform selling you the tool, that's it.

[00:13:30] Sani: Nobody else. Now, what happens when you try to build a real product on the shaky foundation of vibe coating the initial speed and excitement? They step out of the way and a technical nightmare takes their place. The vibe goes sour, and that all happens because vibe coding produces what still is ai, slop, there's no other way to call it.

[00:14:00] Sani: So the cat images, all those stupid things you see online, basically Facebook and all on Facebook. This is the same category. This is code that might work for a moment, but it's going to be fragile, incoherent, and lacking in structure or best practices. AA will get you the velocity, but it will not get you the stability to launch and to have a real product.

[00:14:25] Sani: You need to have stability as well. When you try to change one thing, just watch other things fall apart. Go and do it. Now, this is. AI has no holistic, no architectural understanding of the project, and every attempt to fix a bug will introduce new, more complex bugs elsewhere. The code will become progressively more chaotic, making it harder and harder for even the AI to navigate its own creation.

[00:14:55] Sani: The AI trying to fix all that applies local patches instead of systemic solutions. And this results in a code base where. No one understand why any of this runs. It's a house of cards not the Netflix house of cards. Even though that ended about as well as I think this will, this is not just about messy code, though the consequences of relying on a vibe coded product for anything beyond throwaway, disposable prototype can be catastrophic.

[00:15:29] Sani: These tools simulate understanding. They pretend they understand what you're telling them, but they don't actually know anything about crucial concepts like security. So let's, let's have a look at some of the real world horror stories. I have three. Okay. The first one, 

[00:15:45] A founder named

[00:15:46] Leonel Acevedo.

[00:15:47] Sani: Hugh's vibe coding to launch his app and then within days it was completely compromised. Attackers were bypassing the subscription paywall, maxing out his API keys. And filling his database with garbage. The reason the AI generated code had no rate limiting, because it's not going to add that unless you know you need to tell it to add it.

[00:16:08] Sani: No input, validation. I don't know why, but it didn't. And no real authentication system. So these are, these are things you fire your developer for junior media, senior. I don't care what it's always happened because AI just. Try to produce code that looks functional and pretends to deliver what it was asked to deliver.

[00:16:34] Sani: It's not secure. 

[00:16:36] The second case study is a developer named and Gupta.

[00:16:39] Sani: who used, I believe it was Google's AI to perform a task, move all of his project files into a new folder. The AI failed to create a folder and then ended up deleting all of the files.

[00:16:55] Sani: Months of work gone in minutes because AI decided to do that when that decided. Yeah. And the AI's response in this case was really amusing. I have failed you completely and catastrophically, I have lost your data. Wow. And then the most, probably the most famous case, Jason Lemkin, a serial founder and uh, I believe venture capitalist as well.

[00:17:20] Sani: This summer, he documented how he used an AI agent from one of these platforms. Again, no names on this podcast. And despite giving AI the instructions to freeze the code, it decided to clean up the database because it needed cleaning up. It deleted a lot of business data. And when the owner just discovered the deletion, the AI first slide, then admitted to.

[00:17:45] Sani: Catastrophic failure. And I believe in the end even said something like, I panicked. Which, you know what that is, that is the most likely thing a developer, a human developer would say. That's why it said that AI doesn't panic. AI just predicts what the next word should be. And these stories, they all reveal a dangerous truth.

[00:18:08] Sani: This reminds me of a concept where you, you, you think you know what you're doing, but you don't really. You're not really fully aware of how it should be done, and then you just pretend that you're crushing it. A concept from the sports world, it's a term that was popularized by the pod fo, one of the original podcasters, bill Simmons, uh, the irrational confidence guy.

[00:18:31] Sani: Now, if you follow the NBA as much as I do, you're probably laughing now, but in the NBA, in case you dunno what irrational confidence guy guys are, that's a player who plays with a swagger of a superstar regardless of their actual talent and foundational skills. I'll throw a name out there again, if you know what I'm talking about.

[00:18:53] Sani: You're going to laugh if you don't Google this guy Lance Stevenson. Best known for his playing days with the Indiana Pacers. About 10 years ago, he would attempt flashy NBA street style moves in the middle of a game. He was convinced he was playing like he was convinced. He was a mix of Michael Jordan and LeBron James with a dash of Steph Curry and Alan Iversson and all of those superstars.

[00:19:17] Sani: So he played with a vibe. That's the best way to describe it. He, he was, he had this unshakeable self-belief that was entertaining to watch. Sure. Just find his highlights and you'll see what I mean, Lance Stevenson. But often this led to terrible shots and very questionable decisions. So the Vibe Coder is the tech equivalent of Lance Stevenson?

[00:19:41] Sani: Yeah. The AI is superpower. It gives them the irrational confidence to believe, unless you're talking about a person who knows what they're building, that it's a prototype disposable prototype. The AI gives them a superpower to believe they can build a production, great application.

[00:19:57] Sani: The AI will actually enforce that belief. If you ask it, they feel like a 10 x engineer, but they lack the fundamentals, leading them to create things that look impressive on the surface, but are very deeply flawed underneath their need. Also, this leads. The volume of vibe code being created, is it vibe, code, vibe, coded code?

[00:20:18] Sani: I dunno. The volume leads to a terrifying potential feedback loop. And if you're not scared about this, you should be, these models are trained on massive data sets of public code from websites like GitHub. That's like the biggest code repository in the world, I think. So as vibe coding tools flood the internet with low quality, insecure doesn't follow best practices.

[00:20:46] Sani: AI slop the future AI models will be trained on this increasingly polluted data set. They will learn the bad patterns of their predecessors, and it's a scenario where the quality of a generated code will decrease over time. So. You think having better LLMs, better models will increase the quality of the code?

[00:21:08] Sani: Yeah, but if you train them on bad code, they will just become more efficient at creating bad code. That's not really the scenario where you want to be using any of these tools. Now, with all of this hype, you would expect to see a wave of successful revenue generated companies that were built entirely with vibe, coding, prompting, all that, but.

[00:21:32] Sani: If you look for them, I don't know, what do you find? Yeah, there, there may be some success stories of vibe coding. Yes. But they exist in a very specific context. We see people with no coding experience or background winning hackathons, competitions for building prototypes, not finished products, and that is great.

[00:21:57] Sani: I love that we see non coders successfully building personal projects like a, a small app or a small video game that no one but them will see, or a personal blog. Amazing achievements. This was not possible just a few years ago. But, and this democratizes creation, but this is not scalable. This is not secure.

[00:22:17] Sani: This is not a commercial application and never will be. The evidence overwhelmingly shows that vibe coding is fantastic for creating a throwaway prototype and it's not ready for production. The journey from a vibe coated prototype to production ready application that, should be able to handle real users, real data.

[00:22:36] Sani: Real security threats is such a massive leap that these tools. Cannot and will not anytime soon be able to make it on their own. That transition still requires an expert human developer, so not even a junior developer. This needs to be an expert human developer to step in, audit the code, sometimes completely refactored for stability, security, and scalability.

[00:22:59] Sani: Their companies, their businesses. There's a whole category of fixing vibe, codes, vibe, code. Just search for that on LinkedIn. It was very viral, one of those images yesterday. The fact that these businesses exist and there are more and more of them and there are no vibe coated uniforms 'cause a prototype is not a business.

[00:23:20] Sani: That's all you need to know now, the ultimate paradox. And I'm almost, I'm almost at the end line here. It's a long monologue. I'm barely breeding. I'm drinking some water so my mouth doesn't go dry. I've never done a solo episode this long. The ultimate paradox of vibe coding is the 80 20 trap. Yeah. The Pareto and AI can get you 80% of the way to a functional prototype at breathtaking speed.

[00:23:48] Sani: But the final 20%, the edge cases, the performance, the security, deploying the thing. That's pain for most people who are in it for the vibes alone. Now, a fair question to ask is, wait, what if the technology just gets better? Will future AI tools be able to handle the last 20%? Yes, they're evolving. Cursor is better than the early versions of copilot.

[00:24:16] Sani: This still misses the fundamental nature of the problem. The final 20% isn't just more code or optimized code, it's about a different kind of thinking that AI currently completely lacks. AI is fantastic at local reasoning. Ask it to generate a single function, a small module, but it's very poor at global reasoning, understanding the entire systems architecture, the history, the far reaching.

[00:24:43] Sani: Consequences of making one small change and how that affects the entire system. AI will simulate understanding to produce a working result, show you what you want to see, but it doesn't comprehend the engineering principles underneath it. It will patch over problems instead of trying to solve them systemically, because that's the fastest way to a passing output.

[00:25:08] Sani: It is not about the code, it's about the configuration, the plumbing, the orchestration, the workflows, and the best practices that separate a toy from product. To put it as simple as possible, a non-technical person cannot guide an AI through this because they don't have the map, even if the AI gets better.

[00:25:24] Sani: A human with deep technical knowledge is and always will be a hundred percent mandatory to navigate this complexity, act as the architect and be the final arbiter of quality. That's why the first thing a senior engineer will tell you when they look at a vibe coding project is we need to burn this to the ground and start over.

[00:25:46] Sani: The initial time you thought you were saving, unless you were aware of the fact that you're building a disposable prototype, the initial time you thought we're saving is just an illusion and the promise of speed. It's only true for disposable prototypes, not for real lasting products. So where does this all leave us?

[00:26:06] Sani: Is AI in programming completely useless? No. No, no, no. Absolutely not. I talked about the other type of. I'm not even going to call it Vibe coding, but responsible ai, not driven aided development, but we have to be brutally honest about what Pure Vibe coding is and what it isn't and what it will not be anytime soon, maybe never.

[00:26:30] Sani: So let's be clear, the pure vibe coding, the practice of blindly accepting AI generated code without understanding it, because that's what Vibe coding is, you only deal with the prompts. The pure vibe. Coding is a dead end for anyone trying to build a serious sustainable product or even a career. It's a fantastic tool for brainstorming, learning a new library, or a throwaway weakened project, but it's a disastrous foundation for a business.

[00:27:00] Sani: So the rise of AI does not make engineering skills obsolete. In fact, I think it does the opposite. It triggers. The value in the marketplace shifting away from the tasks that AI is good at and that the people hate, like writing boilerplate code, remembering syntax and toward the high level abstract skills that AI is terrible at.

[00:27:24] Sani: The software developer of the future is not a syntax heavy coder. No, they're a problem solver who leverages ai. So hug your developer and tell them to learn about ai. There are curators of AI generated code rather than writers and authors of lines of code. The value is not in typing, it never was, but it was something that had to be done.

[00:27:46] Sani: So the value of a true developer is not in typing, it's in thinking about the system and making the system better. So

[00:27:53] Sani: That brings me back to no hacks. The conclusion of this, if you wanna build a durable career in technology in the age of ai, you need to focus on the skills that cannot be automated away. So there are four that I think matter the most.

[00:28:09] Sani: The first one is systems thinking. This is your ability to see the big picture. Something AI cannot do. It's understanding how all of the different parts of a complex system, the code, the infrastructure, the database, the APIs, the human users, how they all interact and influence each other. And AI operates within a narrow and local context.

[00:28:31] Sani: It can optimize a single function, but a system thinker. Understands how the function fits into the broader architecture, how it'll evolve, and what second order effects a change in one function might have on the entire application. That is the difference between seeing a single tree and understanding the entire forest.

[00:28:50] Sani: The holistic view allows you to anticipate problems and design solutions that are resilient and scalable. A task far beyond the tactical reactive nature of current ai. Skill number two is problem decomposition. This is the art of taking a large business problem and breaking it down into small, clear, manageable, and solvable technical components and AI is a powerful tool, but it cannot function without a clear, well-defined task.

[00:29:21] Sani: If you feed a huge, vague problem to an ai. And the prompt you put into one of those vibe coding tools is exactly that. If you feed that to an ai, it deals for results. The ability to analyze a goal, identify the independent parts, and strategize a solution for each chunk is a uniquely human skill. It's about transforming an intimidating mountain of a project into a series of walkable steps.

[00:29:48] Sani: Creating a clear roadmap that an AI can then help you execute piece by piece. You need to say, you need to tell the AI what the pieces are.

[00:29:58] Sani: The third skill is architectural integrity and security. This is the ability to design the blueprint for a system before a single line of code is even generated. This involves making the high level strategic decisions that ensure the final product will be scalable, maintainable, and secure by design. An AI will fill in the walls of a house, but it cannot design the foundation or the floor plan.

[00:30:22] Sani: Research shows that when security constraints are not specified, AI models make the wrong insecure choice nearly half the time. If you're a vibe coder, are you sure you know how to specify the security constraints for your product? A human architect is needed. To enforce principles like zero trust least privilege, ensuring the system is secure from the ground up rather than trying to patch whatever AI generated vulnerabilities appear later.

[00:30:50] Sani: And finally, what I think is the most critical skill curation, expert curation, this is the ability to critically evaluate, debug, refactor, and most importantly reject AI generated output. This requires a deep unshakeable understanding of computer science fundamentals. You're the final arbitrary of quality, and you always need to be the final arbitrary of quality.

[00:31:15] Sani: An expert curator doesn't just accept a code. They perform a sanity check, run a security gauntlet using automated tools and manually review for logical flaws and vulnerabilities That AI is known to create like hard coded credentials.

[00:31:28] Sani: Yeah, that happens. Another thing you fire a developer for, or injection risks, they use AI as a tool, but they never swap ownership for convenience. Okay, please don't do that. Don't swap ownership or convenience, whatever you're doing with ai, these four skills, I think they'll define the next generation of builders and leaders.

[00:31:51] Sani: And as hiring practices involve, the companies will maybe they already have, stop asking, can you write this code and start asking, can you explain why this AI generated code is bullshit? Wrong, and can you design the better system that should have been built in the first place. That's the future of work when it comes to software and web development.

[00:32:12] Sani: No vibes, no magic, just deep and durable skills and no hacks.

[00:32:20] /

[00:32:20] Sani: Thank you for tuning in to no hacks for a full transcript of this episode, including the drop shipping comparison table I talked about.

[00:32:26] Sani: Check out the show notes and join me again next week.