After thinking about it for quite a while, I still couldn’t come up with a better title, so I’ll leave it as it is—even though the scope of the topic is obviously very broad. The overwhelming amount of AI-related articles and videos circulating across media platforms has created a sense of anxiety. Personally, this wave of change has already had a noticeable impact on me.
My experience with AI tools evolved gradually. At first, I treated ChatGPT as little more than a conversational toy. Later, I began using coding assistants such as Codex in VS Code. Eventually, I started working directly inside environments integrated with Visual Studio, including TwinCAT 3. The more helpful these tools became, the more anxious I felt. Many skills that engineers once mastered through memorization—or even muscle memory—are rapidly losing their value.
Recently I have been juggling several tasks at once: product design, engineering calculations, PLC project architecture, function block design, and even electrical schematic drafting. At the same time, AI agents are now capable of generating project files or intermediate files directly within local PC directories. That capability alone is already changing the way we work. Inspired by this trend, I decided to write down a few scattered observations and see whether others share similar feelings.
1. The Learning Curve Is Being Compressed
In the past, mastering a technical skill often required a long and painful learning process. Today, AI-based learning assistants can significantly accelerate that process. With proper guidance, a person can reach roughly an “80-point level” in a fraction of the time it used to take. The barrier to entry for many technical skills is being lowered dramatically.
2. The Devaluation of “Human-Memory Skills”
Many of the skills we previously relied on when using tools such as TIA Portal or TwinCAT 3 were essentially manual or memory-based skills.
For example:
-
remembering which library contains a particular function block
-
recalling where a specific operation is located in a menu
-
navigating through layers of configuration dialogs
These abilities are largely based on muscle memory and repetitive practice. Their value is rapidly declining.
Modern development environments such as Visual Studio and VS Code are already integrating mechanisms like MCP (Model Context Protocol), which allow AI to call external tools and services. Similar capabilities are also emerging in industrial engineering software. Interestingly, in one of my conversations with ChatGPT, this capability was humorously described as being “particularly good at clicking buttons.”
3. Many Time-Consuming Tasks Were Never High-Value Work
Tasks such as:
-
batch creation of variables
-
rule-based naming
-
copying standard modules
-
creating similar machine units
-
building the initial project skeleton
used to consume a considerable amount of time in PLC projects. However, the actual technical value of these tasks has always been relatively low. They persisted simply because there was no convenient automation entry point in the past.
Now that AI can perform these operations easily, the inefficiency of doing them manually becomes very obvious.
4. Experience Used to Be Measured by Fragmented Knowledge
In the past, determining whether someone was experienced often depended on their ability to recall fragmented pieces of information quickly. This type of knowledge had value largely because it took other people a long time to search for it.
Unfortunately—or perhaps inevitably—AI systems excel at exactly this type of task: retrieving, organizing, and presenting scattered knowledge almost instantly.
5. Preserving and Reinforcing Personal Value
If we want to maintain and strengthen our professional value, one possible approach is to externalize the knowledge in our heads.
Tools such as Obsidian allow engineers to build structured knowledge systems. It’s hard to say exactly how much value this creates, but documenting and structuring knowledge rarely turns out to be a mistake.
Recently, during a conversation with a friend, he said something that stuck with me:
“It’s not that important to remember everything. What matters more is knowing the constraints.”
This idea seems especially relevant today. The practical value lies in improving our ability to organize context and express technical ideas clearly.
6. The Real Problem with AI Is Responsibility
Since the birth of AI, the main challenge has not been whether AI can generate results—it clearly can. The real issue is: who is responsible for those results?
Looking ahead, I suspect that future versions of engineering platforms such as TIA Portal or TwinCAT will evolve in several directions:
-
stronger Automation Interfaces
-
more powerful openness and integration capabilities
-
IDEs with significantly enhanced verification and validation systems
In other words, development environments may become platforms that emphasize verification, traceability, and responsibility, rather than just code editing.
7. Closed Ecosystems Will Struggle in the AI Era
At a time when AI models are evolving rapidly, everyone hopes to use AI as a productivity multiplier. In such an environment, highly closed PLC ecosystems may face increasing pressure.
The more text-based and transparent project formats become, the easier it is for AI systems to understand, train on, and assist with them.
For example, a proprietary project file with a suffix like .abc is far less AI-friendly than formats such as XML or JSON. Text-based structures allow both humans and AI systems to interpret and manipulate engineering data more efficiently.
