Eventually, you come across a short comment: “I have two cats; overall, it does the job, but the filter is difficult to clean thoroughly.” Finally, something useful. Will it become increasingly difficult to find voices like these on the Internet?

Many words, little experience

Artificial intelligence makes it possible to quickly prepare a product description, guide or answer to a question. For someone running a small business and managing the website alone, that is a considerable convenience. The problem begins when a finished text is published without being checked, followed by another and then another. The website fills up with material that is backed by neither the author’s own knowledge nor a conversation with someone who knows the subject.

Readers do not always need an extensive study. Sometimes they want to know whether they can park near a restaurant or whether the entrance to an office is accessible to someone with a wheelchair. Such information has to be obtained and then updated. Simply arranging sentences efficiently is not enough.

There is also the issue of errors. A government resource devoted to the pitfalls associated with using artificial intelligence draws attention, among other things, to the generation of false information. An answer may sound convincing even if it contains a mistake. A confident tone is not proof that someone has checked the facts.

Who is actually speaking to us?

An author’s byline never guaranteed reliability either. People copied others’ texts, published fabricated opinions and pretended to be experts long before chatbots became widespread. Now, however, preparing many similar materials requires less work.

What is perhaps most troubling is not AI’s presence itself, but the lack of clarity. If a shop informs customers that a machine is responding, it is clear what to expect. It is worse when the shop presents it as a consultant and the customer spends a dozen minutes trying to explain an unusual situation. They receive polite answers, but the matter does not move forward at all.

The same applies to a review. It is acceptable for the author to have improved its language with the help of a program. It is another matter when they describe a week spent in a hotel they never visited. The reader should know whether they are reading an account, an advertisement or a compilation of information from other people.

Privacy is a separate matter

When using new tools, it is easy to focus on what they can do and forget what data we give them. Work documents, private correspondence or information about customers sometimes end up in the chat window. Before pasting them in, it is worth checking the service’s rules and considering whether all the details are necessary.

It is also worth organising our basic knowledge about security. The guide What Is a VPN? explains how an encrypted tunnel works between a device and a VPN server. This connection protection does not mean, however, that data voluntarily entered into a chatbot stops reaching its provider. These are two different matters, although both concern privacy.

A photo no longer ends the discussion

Imagine a photograph of a flooded street published in a local Internet group. Someone adds the name of the town, and another person shares the photo with a warning. Before anyone asks about the author and the date, the material begins to take on a life of its own.

A generated image may look credible, just like an artificially prepared recording. In its materials on recognising deepfakes, NASK shows which elements of an image and sound are worth paying attention to. The mere presence of a face or a familiar voice no longer settles the matter.

In practice, it helps simply to pause before sharing. Who posted the material first? Was the location given? Does anyone who is actually there confirm the information? With news from Bolesławiec, a phone call to a witness may clarify more than a long discussion beneath the photo.

There is another side, too

Not every text supported by AI is empty. Someone with professional knowledge but difficulty writing may finally be able to share it thanks to such a tool. An organiser of a community campaign can prepare an announcement more easily, while someone with a weaker command of Polish can improve a message to neighbours. In such situations, technology helps people communicate with one another.

Much therefore depends on where the program’s work ends and human responsibility begins. The author should read the text, check the information and put their name to what they publish. Just as before, only the temptation to skip these steps is greater.

The Internet will remain a place for people to meet if it is still possible to find someone’s own experience, an honest answer and a willingness to correct a mistake there. Even a short comment about a vacuum cleaner’s filter then has more value than another flawlessly written guide that says nothing of substance.