Today, much speculation exists about how AI will impact the job market and how revolutionary and disruptive it will be. In this article, I want to discuss how to use AI in software development today, not in the future.
Use with caution
AI, as we know it today, is mainly machine learning. AI currently is just statistics.
Machine learning, in short, is statistics, just a large amount of data that the computer “learns on it”. Given that it is just math, the computer tries to give you his best guess. The best guess is correct most of the time, but what is the best guess? The highest probability might be incorrect, so you must be careful. AI will either be a genius or a complete Pinoquio.
Right now, the best use for AI is a supervised co-pilot, in which you double-check its answers.
AI use cases
Multiple tools are available, and new ones appear daily, each with a specialised use. Here are the AI tools I use during each part of the development phase. I use Chat GPT and Claude for LLMs and often use them as 2 co-pilots if I need two opinions.
