AI Tools for Mobile Development: Pros and Cons of using AI Assistants

This is the second part of my personal experience using AI tools for mobile development. (Read the first part here) In this article, I want to share the advantages and disadvantages of using AI tools in Mobile development and discuss my overall experience. As always, the article picture is AI-generated, but my opinion is not. 

Let’s start with the positive aspects of using AI tools.

After an initial struggle to understand how you, as a developer, can communicate with AI, it will help speed up your workflow. Mundane tasks, UI tests, or documentation are less of a struggle when AI is your coding buddy. 

AI is a great companion if you are a solo developer on a small project. AI can handle code reviews, help with choosing approaches to new features, and is great with easy bugfixes. When there is no one to help, AI is the best coding buddy for you (after rubber ducks, of course).

Also, AI is very helpful when you hit days of “coding slump” or other struggles of some sort. We all, humans, have our bad and good days, so AI helps a lot when you struggle to start a new task or need to “just do it”. You feel a lot more confident, deadlines become less stressful points, and it improves code quality overall.

However, there are downsides to using AI in development.

First and foremost, AI is making mistakes. And it will do this all the time, so you need to be careful about what AI does in your project and check every commit thoroughly. I’ve had a few times in my practice when AI deleted chunks of code without asking for it, added bugs by writing easier implementations, or just generated bad quality code. So sorry, but “vibe coding” is not an option at the moment for serious projects.

Another AI problem that I experience is that AI lacks context. It is natural, because AI can only have your prompt, codebase, and its own knowledge, and that’s all. But sometimes it will create really deep bugs, and it can bring major damage to a project. So I can’t recommend using AI at full speed on sensitive data or some core logic. Be careful about this.

And last one, AI tries to generate a code that will compile and be as close to the prompt as possible, but AI surely will ignore codestyle in most cases. Now it is a lesser issue since we have an AI plugin for your IDE, and you can add rules or add a mandatory part of a prompt to use code style guidelines, but it is still an issue. Also, AI rarely tries to criticize your code properly or say that your approach is wrong for your technology; it will try to adapt to what you ask it to do. So make sure that you know what you want to get, or it will be an unpredictable piece of code.

I have to say that AI is here to stay and most developers need to use it.

AI positively impacts your workflow, helping with a lot of tasks and improving estimates. Nowadays, most clients will request you to use AI because time is always crucial for business. But still, it is only you who will be answerable for your codebase, so I ask you to be cautious and keep your code clean. This way, you will have a balance between productivity from AI and quality that will suit yours and your client’s standards.

Swan Software Solutions is committed to creating reliable, scalable, and affordable solutions. To discover more about how we can help your company implement AI or for other technology needs, schedule a free assessment.