When I did a summary of a website that I was doing a quick market research for a client and it recommended it was actually a co-pilot AI on my browser for Microsoft edge and it suggested that this company uses a Microsoft product and when I ask them during an interview they actually said no we don't use anything Microsoft product. So there's a bit of a product placement within their AI checkbook to summarize a website that was false basically.
All the photos and memes that are made today of kittens, for example, may seem completely harmless and purely for entertainment, but they are also part of the massive amount of data that exists online and can be used to train or improve AI systems. Every image, caption, and interaction contributes to patterns that AI can learn from—such as recognizing what a kitten looks like, understanding humor, or identifying what kind of content people engage with the most. Because of this, even lighthearted content plays a role in shaping how AI systems “see” the world and respond to users. While a single meme doesn’t have any noticeable impact on its own, the accumulation of millions of similar pieces of content helps build datasets that influence how accurate, creative, or biased an AI system can become. This highlights an important idea: AI is not only trained on formal or scientific data, but also on everyday digital Culture.
Yes, actually, just today. We were working on a rather extensive project in which we needed to automate a part related to business sectors. We asked artificial intelligence to extract the sectors of the Basque Country from a very large table. The problem was that it made a mistake, and it was quite an obvious one—it left out some important sectors, and we weren’t entirely sure why. At that moment, we realized that we couldn’t trust the result, so we decided to do it manually. Of course, it was a bit frustrating, because you invest time in carefully formulating the question and trying to optimize the process, and yet it still doesn’t work correctly. But that’s precisely why it’s so important to always review the outputs and not rely on them blindly. Taking the time to verify and correct the AI’s work ensures that mistakes don’t go unnoticed and helps maintain the accuracy and reliability of your results.
A concrete example from my work in the field of project coordination: I asked an AI to analyze two consulting services for leadership development – including a comparison of advantages and disadvantages as well as a cost-benefit calculation. The AI delivered a visually convincing, structured analysis with a clear table display. On first reading, everything seemed conclusive – structure, language and logic were right. It was only when I specifically checked the figures that I realized that the derivation of the consultation days was simply wrong. The AI had not correctly interpreted the information in the original documents and used it to construct an erroneous calculation – which, however, did not immediately catch the eye in the overall picture. What stopped me? Firstly, my principle of always checking figures yourself – especially when it comes to decision-making bases with financial consequences. Second, the knowledge that AI does not 'understand' original documents, but recognizes and extrapolates patterns. Since I had the offers myself, I was able to compare directly. Without this direct comparison, the error would probably have gone unnoticed. The lesson from this is that AI is an excellent first instance for structure and formulation, but not a reliable computational authority for subject-specific derivations. This example impressively shows why a structured plausibility check remains irreplaceable in knowledge management and project coordination. The human expert view – with contextual knowledge and industry experience – is the decisive corrective in AI-supported processes.
I was supposed to interview an athlete – Reinfried Herbst, an Austrian ski racer. The AI provided me with a summary of his career as preparation. Something about it seemed strange to me – a detail didn't feel right. I checked, and indeed: the information was wrong. If I had gone into the interview with this information, it would have been embarrassing – towards the athlete and on camera. What stopped me? Not a system, but gut feeling and the habit of questioning facts
I rely on it mostly for my daily work and not really a lot for daily-life scenarios, so describing exactly how suggestions are wrong would make little sense. That said, I can confidently say that it makes quite a lot of bad suggestions, mostly born from a lack of context, but it is way more than one would expect. It has definitely improved in the last iterations, but as a rule of thumb, I always distrust what it suggests until proven otherwise; a mentality that can only be born from having a big amount of disappointments.
Misjudgments regarding health conditions, misunderstanding the question, or a programming error in the generated code. I was stumped by the basic logical analysis of the text, which is a mandatory step when responding to AI. Furthermore, you need to understand the subject matter you’re asking about. AI IS A HELPER, NOT A REPLACEMENT!!!
In the last 24 hours, algorithms have supported me in multiple small but important ways, often without me noticing at first. For example, when I checked social media, the content I saw was already filtered and prioritized based on my interests and behavior. This saved me time and helped me focus on relevant information. In my professional work, AI played a key role in researching destinations, doing quick and efficient analysis of data, and preparing presentations. It also helped me summarize tasks and provided practical suggestions that improved my decision-making and productivity. When I used maps or checked traffic conditions, algorithms optimized my route in real time. Even email filtering and spam detection quietly improved my workflow by organizing information and reducing noise. What is interesting is that most of these interactions feel natural and almost invisible. We don’t actively think about the algorithms, but they shape many of our daily decisions and significantly increase our efficiency.