AI has become part of how we write, research, design, and strategize. The question is no longer whether to use it, because there is dependence on AI, and increasingly, we probably should use it. The deeper question is what happens when using AI becomes easier than thinking the problem through ourselves.
I notice AI overuse in something as ordinary as email. Someone writes a message, reads it twice, knows exactly what they mean, then still sends it to an AI tool for approval before hitting send. I do it too, sometimes. It’s like running a calculator to confirm that an answer we already know is right. The tool is useful, but after a while, it raises a more uncomfortable question: are we simply double-checking our work, or slowly losing confidence in our own judgment?
That critical thinking and AI distinction matters, because not everyone uses AI the same way. Some people use it as they once used Google, books or research databases: to find information, compare ideas and understand a subject faster. Others are beginning to use it as a substitute for the process itself.
A 2025 Microsoft Research study and Carnergie Mellon University, surveying 319 knowledge workers, found that higher confidence in generative AI was linked to less reported critical-thinking effort, while higher confidence in one’s own abilities was linked to more. The researchers also found that AI dependency tends to shift critical thinking toward verifying information, integrating responses and overseeing the task instead of performing it directly.
That does not mean AI is automatically making people think less. But it does suggest that our relationship with the tool matters. There’s a real difference between using AI to challenge your thinking and using it so you don’t have to think as much in the first place.
Prompting’s Not Strategy
I keep seeing courses built around prompting, as if finding the perfect sentence to type into a chatbot can somehow replace expertise. This becomes even more obvious in professional work.
Prompting is useful, no one is denying that. Giving AI the right context, asking the right questions and knowing how to refine an output can make a huge difference. But a good prompt cannot rescue a weak brief.
In marketing, for example, an experienced professional knows which questions have to come before a strategy: Who is the audience? What problem are we trying to solve? What’s the positioning? What are the business objectives? What are the limitations? What is the competition doing, and why should anyone choose us instead?
AI can help research the market, organize information, challenge assumptions, build content pipelines and speed up execution. I use it for exactly these kinds of tasks. But asking “give me a marketing strategy” without that foundation can easily produce something polished, structured and convincing that still says very little.
That is what concerns me, dependence on AI. AI is becoming exceptionally good at producing work that looks complete, even when the thinking behind it is not.
When Design Becomes a Prompt
The same over reliance AI tension is becoming visible in design.
Universities still teach Photoshop, Illustrator, 3ds Max and AutoCAD while AI becomes increasingly embedded inside those creative workflows. Adobe Illustrator can already generate fully editable vector graphics from text prompts, according to Adobe’s official Illustrator AI documentation.
That does not mean Illustrator is disappearing, and it certainly does not mean design education is becoming irrelevant. If anything, it could make the foundations of design even more important.
When execution becomes easier, understanding typography, composition, hierarchy, production and concept development matters more, not less. The real value of a designer has never been simply knowing where to place a button. It’s understanding why something works, why another option doesn’t, and what should change before the work is considered finished.
The question becomes more interesting when AI can generate not just an image, but a complete, editable production file. At that point, knowing which software command to use may matter less. Knowing whether the result is technically correct, visually strong and appropriate for its purpose will still require judgment.
China offers an interesting signal, although the story is broader than graphic design. In 2025, China’s Ministry of Education added 29 undergraduate majors to its national catalogue. Among them were AI education, intelligent audio-visual engineering and digital drama, alongside other programs responding to technological, market and national priorities.
What I find important here is not that traditional creative disciplines are disappearing. It is that education is beginning to expand around combinations of technology and existing fields. That is probably where design education has to go as well.
What About Generations Raised on AI?
This is the part of the conversation I find most important.
People who learned to research, write, design or develop strategies before generative AI built those foundations without it. They made mistakes, struggled with tools, searched for answers, rewrote sentences, and learned to recognize when something simply didn’t work. AI arrived later and became another layer on top of that experience.
The next generation may not have the same dependence on AI.
They may encounter AI before they have developed the confidence to form an argument, solve a problem or defend a creative decision independently. That does not make them less capable. But it does mean schools and universities have a responsibility to think carefully about what students are still being asked to learn for themselves.
Banning AI is not the answer. Education has to adapt to it.
England’s Department for Education recommends that AI products for learners mitigate the risk of cognitive deskilling and avoid providing complete answers by default. Instead, the guidance encourages systems to ask learners to attempt problems first and progressively provide support.
That feels much closer to the right direction. AI should help students reach an answer, question it and improve it. It should not become a shortcut around the learning process itself.
When AI Can Do Even More
This debate will only become more urgent as AI systems grow more capable.
OpenAI’s GPT-6 Astra, announced in September 2026, extends AI further into research, computer use and complex multi-step work. As these systems become better at executing entire workflows, more tasks that currently require professional intervention may become automated.
That does not necessarily make human judgment less valuable. In many cases, it may make judgment the most valuable part left.
I am not arguing that people should use AI less. I think we should use it, learn it and adapt our jobs around it. It can make us faster, expose us to ideas we might have missed and remove hours of repetitive work. Pretending otherwise would be unrealistic.
What I do worry about is reaching a point where the presence of a good answer makes us stop asking whether we understand it.
The challenge ahead is not simply learning how to prompt better or how to use the newest AI model. It is preserving the ability to question an answer, defend a decision, and recognize when something that sounds intelligent is actually wrong.
AI can give us more answers than we have ever had before. What will matter is whether we are still building the judgment needed to know which ones deserve our trust.
Inside Telecom provides you with an extensive list of content covering all aspects of the tech industry. Keep an eye on our Opinion sections to stay informed and up-to-date with our daily articles.
