AI supports as a tool. We humans decide.
AI Is a Tool and Not a Replacement
AI is currently on everyone’s lips. And as is often the case, 100 people have 100 different opinions on it. Ranging from „useless“ and „practical“ to „job killer“ and „the beginning of the end,“ pretty much everything is represented. At HeiReS, we are also repeatedly asked whether AI could help with topics like accessibility, for example.
Of course, we use AI at HeiReS. But in the way we believe is right—namely, as a tool for specific tasks. We also engage intensively with the EU AI Act and, in connection with it, the labeling requirement for AI-generated content. There is no question that AI and its usage should be monitored and communicated transparently. However, for such a label to be sufficiently understandable, it must also reflect how AI is actually used.
The title of this post says it all: for us, AI is a tool. And like any tool, AI serves to accomplish specific tasks where it can be sensibly deployed. The emphasis here is on „sensibly.“ And just like with all tools, AI must be used with professional expertise and clear purpose.
Humans Decide on Quality
AI is a panacea for nothing! Almost nothing that comes out of an AI can be used in an uncontrolled or uncorrected state. And the better you know how to operate a given AI, the better the results will match what you expected. This means the actual performance is delivered by the human who tells the AI what to do, and this must be constantly monitored and corrected.
So-called „prompt engineering“—formulating the requirements of what the AI should do as precisely, purposefully, and appropriately as possible—is almost a science in itself. But dealing with it is essential if AI results are not to be mere random hits that only roughly match what you actually want. Precise results require precise phrasing… and sometimes a great deal of time.
AI Has Limits
Like any tool, AI has limits to its applicability. Even though AI can now „speak“ and act to some degree like a human, it will never truly think and feel like a human. That is precisely the crux of the matter when it comes to usability and accessibility, for example. Of course, purely technical aspects can be checked and potentially corrected using AI. But in everything where you literally have to put yourself in another person’s shoes, an AI can only imitate, not empathize.
Proper Usage Makes the Difference
If you consider: 1. exact and precise prompting, 2. using the right AI suited to the task, 3. realistic expectations, and 4. responsible handling of the results—then AI can be an outstanding, time-saving tool that delivers quality outcomes. But truly only then.
Transparency Instead of AI Marketing
That is why we will not mark content developed with the help of AI using a microchip, a robot, or anything similar. Because that simply does not do justice to the partial human effort involved. A carpenter wouldn’t stamp „Built with hammer and saw“ onto their craftsmanship either. Our labeling always clearly states that things are created by humans using AI as a tool. That is the only way AI should be used.
Article by HeiReS employee Armin Reuter – may contain individual opinions
