Web programming in the AI era and the future of the job

TL;DR: If you're feeling too lazy to read this post, get yourself Digestmark, a browser extension I built and summarize it 😉.

Yesterday, I was invited to speak to software engineering students about the impact of AI on web development and the future of our profession. The invitation forced me to step back from my day-to-day development work and really examine this question that's been circulating through tech communities since ChatGPT's release: Will AI replace developers?

As someone who's been building web applications for years, I realized I had developed strong opinions about AI tools, both positive and skeptical, based on my personal experience. But speaking to students who are just beginning their careers meant I needed to move beyond my own biases and look at the data, the trends, and the actual capabilities of current AI systems objectively.

Here's what I shared with those students about the impact AI is having on our field.

Some context about the AI revolution

We've been trying to make intelligent computers for more than 50 years. A lot of them excelled at precise industrial tasks, character recognition, speech processing or chess / go games etc., for a while. The breakthrough we witnessed around 2020 was the convergence of three crucial factors:

1. Massive data availability: Decades of internet content provided unprecedented training material
2. Computational power: Cloud infrastructure finally made large-scale training feasible and trained models accessible to the large public at low cost
3. Algorithmic advances: Transformer architectures unlocked new possibilities for natural language processing

This wasn't an overnight revolution and AI has been here for a long time. What changed was the ability to create general-purpose AI that could understand and generate human language across virtually any domain.

The global AI landscape

The competition is fierce and international:

• United States: Leading with OpenAI (ChatGPT), Anthropic (Claude), and Google (Gemini)
• China: Rapidly advancing with companies like DeepSeek, Baidu, and Alibaba
• Global acceleration: Each breakthrough triggers rapid iteration across the entire ecosystem

This competitive environment is driving innovation at an unprecedented pace, with new capabilities emerging quickly.

---

I think it makes sense to divide AI use in web development, or development in general, in two different approaches.

"Vibe Coding" - Democratizing app creation

Andrej Karpathy, Tesla's former AI director, coined the term "vibe coding" to describe a phenomenon I've been observing: people with no programming background successfully building applications using AI.

The Process:

• Natural language description of what you want
• AI generates complete, working code
• Minimal technical knowledge required

The Tools Enabling This:

• Lovable, V0, Bolt.new, Anything etc
• Claude Artifacts, ChatGPT's custom GPTs

This democratization is real and significant. People are solving actual problems with software they create themselves. But still, there are important limitations:

• Complexity ceiling: These tools are great for simple to moderately complex applications but struggle with large-scale, complex systems
• Quality and security: AI-generated code often requires significant review and refinement to meet production standards
• Maintenance challenges: Ongoing updates and bug fixes can be difficult
• Limited customization: Tailoring solutions to specific business needs can be challenging

Professional enhancement - AI as a force multiplier

In professional development environments, AI is being used very differently:

• Intelligent autocompletion: Context-aware suggestions that understand your entire codebase
• Code generation: Creating boilerplate and implementing well-defined features
• Debugging assistance: Identifying and explaining errors
• Documentation and Tests: Auto-generating comments and API docs as well as software tests
• Refactoring: Suggesting improvements to existing code

Industry Adoption: Companies like Shopify have made AI tool usage mandatory for developers, some others are reporting 30-50% productivity increases. But these aren't replacing developers, they're making existing developers significantly more effective.

That kind of AI usage happens in well supervised environments with always a human review down the line before the code is shipped to production.

The impact

As you can see, AI is impacting web development and I think that's both positive and negative in some ways.

Positive effects

• Non-technical founders can build MVPs
• Faster prototyping and validation
• Lower barrier to entry for learning programming
• Faster development cycles
• Ability to experiment with new technologies quickly
• Smaller teams can tackle larger projects
• APIs enabling more complex applications, integrating AI at their core
• More focus on high-level design and architecture
• Increased productivity for experienced developers

Negative effects

• Over-reliance on AI can lead to skill degradation, loss of fundamental understanding and problem-solving skills
• Quality and security concerns with AI-generated code
• Potential for increased technical debt if AI-generated code isn't properly reviewed
• Hype cycles leading to unrealistic expectations
• There are still questions about the real productivity provided by AI tools. Recent software engineering studies suggest that while developers feel more productive using AI assistants, the actual delivered value often falls short of expectations. The time saved on initial code generation is frequently offset by increased debugging, code review, and maintenance overhead.

I think it all boils down to the fact that AI is a tool, and like any tool, its value depends on how effectively it's used. It's not a magic wand that will solve all problems or replace human ingenuity, at least not yet. It's important to keep a deep understanding of the fundamentals of development, architecture, security and performance while leveraging AI to enhance productivity.

The Skills That Actually Matter

Code Review and Quality Assurance: AI-generated code needs human oversight. The ability to quickly identify security vulnerabilities, performance issues, and maintainability problems is becoming a core skill and one can only do that with a deep understanding of the fundamentals.

Problem Decomposition: AI excels at implementing well-defined features but struggles with breaking complex business requirements into manageable technical tasks.

System Architecture and Design: As AI handles more implementation details, the ability to design scalable, secure systems becomes more valuable.

Project Management and Resource Planning: Faster development means more projects, more stakeholders, and more coordination challenges. Strong organizational and communication skills are essential. As as developer, you also need to understand the capabilities of different AI tools and coordinate them effectively, to get the best results.

Separating hype from reality

There's a lot of hype online, majorly from VCs, AI startups and their leaders about AI replacing all developers in a near future. While there's some truth about that, I think we're still far from seeing AI produce the expected standard. We may get there in the future but until then, here is what I think is the important things to focus on as a developer :

• Go beyond the basics: If the only thing that you know how to do is basic UI and backend applications, copy-pasting code from the internet without deep understanding I think you're already replaced by no-code tools a decade ago. AI isn't your biggest threat, your lack of depth is. Developers who only work at the surface level have always been vulnerable to automation, whether it's WordPress themes, drag-and-drop builders, or now AI code generators.
• Stay adaptable: I think if there's only one thing that you need to avoid, both about AI and any other online hype-drama, is extremism. You don't always have to choose. Most of the time, the truth lies in between. Don't accept statements like AI is going to replace all developers soon or AI is just a hype, embrace the evolution, test new things, stay up to day and see how those tools can help you learn faster and do more, more quickly.

I think if AI becomes one day able to do what actual developers do right now, with a high quality, the concerns will be bigger than it replacing developers because that means it will be able to replace most of the office workers today, including lawyers, financial analysts, marketers etc and at that moment, we'll have to answer bigger questions like what's our future as humans. I wrote some thoughts about that recently.

FAQs

WIP, actual questions asked during the presentation.