Artificial intelligence is making a bigger splash in software development than the old fourth-generation languages (4GL) ever did. But if you look closely, some similarities between AI tools and 4GLs are pretty clear. It’s worth exploring whether AI is really the new 4GL or if it’s just another step in the long road of software evolution.

What Are 4GLs Anyway?

Fourth-generation languages, or 4GLs, were supposed to change everything. They aimed to make programming easier by allowing developers to write more natural, human-like code. Instead of typing out complex commands, users could describe what they wanted in plain language or use visual tools. Think of them as program-generating languages that let you build applications faster. This idea was exciting, especially back in the early 1980s, when books predicted AI would someday replace human programmers completely.

However, the reality turned out differently. Actual 4GLs like FOCUS, along with modern tools like WYSIWYG editors, rapid application development frameworks, and low-code/no-code platforms, all still need someone who understands how to work with them. They don’t eliminate the need for skilled developers. You can’t just describe your app and walk away. There’s always a need for someone who knows how to fix issues or make adjustments when things go wrong.

AI and 4GLs: Similar Promises, Different Outcomes

Some people see AI as a kind of 4GL on steroids. It can understand broad instructions and generate code that seems to do what you want. But in practice, AI isn’t quite there yet. Giving AI a vague description of a task often leads to incomplete or incorrect results. The best way to work with AI today involves a back-and-forth process. You refine your instructions, review the output, and tweak as needed. It’s a collaborative dance between human and machine.

What’s clear is that AI tools excel at prototyping and handling simple tasks. But when it comes to building complex, reliable software, a human’s understanding of the details still matters a lot. Abstractions can hide complexity, but they can’t replace deep knowledge. When you’re working on the final stages of a product—making sure everything works smoothly and securely—you need someone who truly understands the system. That’s the human trait of “will to completion,” which AI can’t replicate.

Humans Bring Care and Judgment to Coding

AI doesn’t care about the end goal. It confidently produces code that might be wrong or break other parts of your system. Experienced developers know this and can quickly spot errors or fix broken code. But less experienced programmers might rely too heavily on AI and not catch mistakes. This can lead to what’s called “comprehension debt”—when you use tools that mask complexity but don’t actually teach you how things work underneath.

A good example from everyday life is CSS styling. It’s not my strong suit, and I once thought AI could help me get better at it. Instead, I just ended up with AI-generated CSS that still needed fixing. The real fix was working with someone who understands CSS inside out. Human expertise, patience, and a willingness to see a project through are irreplaceable. AI can assist, but it can’t drive the process or truly care about the outcome.

In the end, the dream of AI is similar to that of 4GLs—making programming easier and more accessible. But both have limitations. AI has become more influential in practice than 4GL ever was, yet it still requires skilled people to make it work effectively. Without that human touch, AI-driven code can lead to more confusion and technical debt. So, while AI is a powerful tool, it isn’t a substitute for experienced programmers who care about their work and see projects through to the end.

Inspired by

• https://www.infoworld.com/article/4049946/is-ai-the-4gl-weve-been-waiting-for.html

Sources

• gmpg.org
• pexels.com
• wikipedia.org
• archive.org
• podigee.com