Rakia Sassi dug into the good, the bad, and yes, the ugly of AI’s role in software development. The reason is that while there is a lot of buzz about AI making developers’ jobs easier, there is more to this story than the glamorous headlines suggest. Let’s get right into it!

The Good: AI Can Boost Efficiency But at a Cost

AI undeniably brings some impressive tools to the table. Generative models like GitHub Copilot and ChatGPT—especially its recent models like the o1-preview—have become real assets for developers. You can ask AI to generate snippets of code or assist in optimizing algorithms, which saves a ton of time on repetitive tasks.

As Martin Heller showed in his article “ChatGPT o1-preview excels at code generation,” the model generated multiple versions of a C++ QuickSort algorithm and improved each time with more efficient features like randomized pivot selection, parallelization, and even adding test suites.

AI can handle simple coding tasks and suggest optimizations that might take a junior developer hours or even days to complete. This allows us, as developers, to stay “in the zone” and spend more time focusing on solving more complex, creative problems.

Even-grade tools are joining the AI revolution. Google’s Gemini Code Assist Enterprise, for example, takes coding to a new level by offering contextual suggestions across the entire tech stack while maintaining enterprise-grade security. There is a lot that this seamless integration can offer:

• helps developers not only to write code faster but also to understand a broader set of services without having to dive into every documentation page
• helps also to manage infrastructure, optimize resources, and even deal with mundane tasks like testing and debugging faster

The Catch

AI tools come at a price, and that price is not just monetary. In many cases, as these systems handle more of the coding process, developers risk becoming overreliant on AI-generated solutions.

As helpful as these tools are, they can lull developers into a false sense of security and prevent them from diving deep into the code to understand what’s happening behind the scenes.

AI does not replace a developer’s need to think critically—it just assists in executing tasks more efficiently.

The Bad: AI Still Struggles With Complex Problems

AI may be good at tackling repetitive and standard coding tasks, but it hits a wall when it comes to more creative, complex, or architecture-heavy decisions.

While it can optimize a QuickSort algorithm for you, it still needs nudging and guidance from the user to reach an optimal solution. In other words, AI is great for refining what’s already known, but it struggles to generate novel solutions without human intervention.

And do not forget that AI-generated code is highly dependent on the data it was trained on, which can result in suboptimal or insecure code if used thoughtlessly. Sure, it can piece together a working solution, but it doesn’t necessarily consider the intricacies of the project you’re working on.

AI lacks context and the ability to understand business logic fully.

Security flaws, performance bottlenecks, and other bugs could slip through because AI doesn’t fully understand the environment in which the code is running.

Case Study: ChatGPT vs. Human Programmer

Let’s take a simple challenge: extracting fields from a text using specific delimiters.

In the following video, I, as a seasoned human programmer, went head-to-head with ChatGPT to implement the code that achieved that goal.

What unfolded was not just a mere coding showdown. It was a battle of wits and strategy that reveals the starkly different ways AI and human minds dissect and conquer problems:

ChatGPT relies on vast data and algorithms and processes information at a high speed. On the other hand, I rely on experience, intuition, and my spark of creativity.

From the efficiency of the code to its maintainability and clarity, can you guess who got it right?

Here is the result:

As you can see, AI is good at mimicking what it’s seen before—great for routine tasks—but for more complex, innovative applications, human intuition and experience are still irreplaceable.

The Ugly: AI Could Intensify the Power Gap in Tech

Here’s where things get a little more concerning: AI’s benefits, like those of cloud computing, are not equally distributed.

In an ideal world, AI would democratize software development and give every developer, no matter their resources, access to the same high-powered tools. But, as Matt Asay in his piece “Open source isn’t going to save AI” rightly pointed out, AI is heading in the same direction as other technological advances: into the hands of a few powerful companies.

Even though open-source AI tools exist, companies like Google, Microsoft, and Meta are the ones with the capital to make AI usable at scale.

Take Databricks Apps, for example. It promises businesses will be able to create custom, AI-powered apps in just minutes. But here’s the catch—this type of speed and efficiency is possible because companies like Databricks handle all the heavy lifting in the background, including serverless computing, security, and data governance.

Smaller companies, or individual developers, increasingly rely on these giants to deliver convenience and simplicity.

The result?

AI is not democratizing development—it is consolidating power in the hands of a few big players, much like what happened with cloud computing.

As Matt Asay argues, while open-source tools do offer opportunities, they often lead to more complexity, which users then pay large companies to manage.

This increasing reliance on corporate-backed AI tools means that unless you’re working for one of these companies or can afford their services, you may find yourself left behind.

The Human Touch: Why You Still Matter?

Take security, for example. AI tools like Gemini Code Assist Enterprise promise enterprise-grade security features, but that doesn’t mean you can blindly trust it to handle all vulnerabilities. As David Linthicum points out, AI-driven code generation can leave security vulnerabilities that remain hidden until a breach happens.

In this hands-on tutorial, which is the third part of my Kubernetes series (here are part 1 & part 2), you can learn how to deploy a Spring Boot app on Kubernetes. The application needs a PostgreSQL database, so you will see how to:

• Use persistent storage by leveraging Kubernetes Persistent Volume (PV) and Persistent Volume Claim (PVC)
• Expose the spring boot app and PostgreSQL using Kubernetes NodePort service
• Access the database from outside the Kubernetes cluster

Click here to read more on why you matter.