DEV Community

Cover image for Developer Diary #27. AI Trap
Kiolk
Kiolk

Posted on

Developer Diary #27. AI Trap

Over the past year, I’ve tried different tools based on AI. I used it for code review in pull requests. I explored some IDE tools and used AI for code generation by prompting. Of course, I also experimented with several chatbots from the major players in the market. I found it quite enjoyable, especially at the start when you prompt and get results almost immediately. It reminded me of the excitement I felt when I first created my own “Hello world” program, though I also remember what came next. I had several pet projects with messy of spaghetti code, where all functionality was placed in one class. I think AI models perform better than I did at my first attempt, but it’s still so painful to understand such code. It becomes doubly complicated when working with unfamiliar technology. I know the approach to understanding; I should go back to the beginning and proceed step by step, but it feels like writing the program by yourself.
They asked an AI engine to recreate “a salmon swimming down a river”, and here’s what it guessed it would look like. Tread in Reddit.They asked an AI engine to recreate “a salmon swimming down a river”, and here’s what it guessed it would look like. Tread in Reddit

Another issue with these tools is their lack of accuracy. When you write code regularly, you do this through iteration. Any problems with code appear during development, such as compilation errors or crashes during execution, etc. However, generated code mostly works, but the big question is validating this work. You should properly test it. Unfortunately, you can find many incorrect pieces of data at this step.

Personally, the efficiency of these tools lies in solving small, concrete tasks that can be simply described in one or two sentences when you have a clear vision of what you expect to get as a result. It is important for quick validation of output. These tools are extremely effective for solving boring and typical tasks, such as creating a mapper or model base on JSON, etc. I like to use it as an alternative to searching and googling. Now it is a new standard for looking for information; you don’t need to spend a lot of time trying to extract information from different resources. Sometimes, when someone asks me for help and the answer is very clear, I want to ask: “Did you try to find the answer in ChatGPT?”
Cool service, Let Me Google That For YouCool service, Let Me Google That For You
AI tools are very powerful, but we should avoid becoming dependent on them. The value of our existing experience is increased, and we should care more about our skills in troubleshooting and problem-solving. Stay focused and don’t fall into the AI trap.

You can find me in X, GitHub, medium or LinkedIn. Thanks for your time and see you in next post.

Image of Datadog

The Future of AI, LLMs, and Observability on Google Cloud

Datadog sat down with Google’s Director of AI to discuss the current and future states of AI, ML, and LLMs on Google Cloud. Discover 7 key insights for technical leaders, covering everything from upskilling teams to observability best practices

Learn More

Top comments (0)

Qodo Takeover

Introducing Qodo Gen 1.0: Transform Your Workflow with Agentic AI

Rather than just generating snippets, our agents understand your entire project context, can make decisions, use tools, and carry out tasks autonomously.

Read full post

👋 Kindness is contagious

Please leave a ❤️ or a friendly comment on this post if you found it helpful!

Okay