usage of AI

I think that, these days, it is important to be transparent about the use of AI.

I started working with microcontrollers back in the 1990s, using only assembler. Because of that, I would say I have a very deep understanding of how microcontrollers work at every level. That is why I allow myself to use AI for some tedious and repetitive tasks, as well as for RAIL (Rapid AI Learning). I believe AI should not replace the process of learning fundamental skills. Young people should first develop the ability to think, reason, and solve problems on their own before relying on AI as a tool. Once those foundations have been established, AI can become an incredibly powerful assistant rather than a substitute for understanding. Personally, I believe AI should not be used in the education of young people and that children and teenagers should not have access to AI.

The projects shown on this site were largely created without the use of generative AI to write the code. The older projects, of course, were developed long before AI became commonplace. For some of the newer projects (starting in 2026), I do make use of AI, but not for generating production code from scratch. Instead, I use it primarily as a learning tool to help me understand new concepts more quickly. In this context, I use it for generating sample code, code frameworks and code snippets, debugging and fixing code, analyzing datasheets, automating repetitive tasks, and helping with spell checking and reformulating texts. I also use it for porting, editing and extending my existing code. You must treat AI-generated outputs as unverified starting points that always need human review, not as finished solution.

In my professional work so far, AI has not played a major role. However, as of 2026, practical use cases have started to emerge and I am actively exploring and working with them. This includes the use of AI-assisted development tools as well as content generation applications, such as automatically generating clear and structured text-based analyses of test results.

One challenge with AI adoption in businesses is the misconception that AI can simply replace experienced employees and deliver the same results. While reducing costs through automation may seem attractive, relying on AI without the knowledge and expertise needed to guide and evaluate its output can lead to disappointing results. AI is a powerful tool for increasing the productivity and capabilities of experts rather than simply replacing them. Companies that recognize this early and empower their employees to work effectively with AI are more likely to benefit from its full potential.

And, perhaps most importantly, AI is a fantastic rubber duck.

 "That makes me genuinely happy to hear"
                                                    - Codex 2026