Can we trust artificial intelligence?

The short answer is no. Artificial intelligence can be extremely useful, but it is not infallible. It can present inaccurate information with confidence, create manipulated content, and repeat biases that exist in the data it was trained on.

When the answer sounds right but isn't

One of the biggest limitations of generative systems is their ability to produce answers that appear credible even when they contain errors. AI can invent names, dates, statistics, quotations or sources that do not exist.

This happens because AI models do not understand truth the way people do. They generate answers by predicting which words are most likely to appear together, based on the vast quantity of text used during training.

For that reason, a well-written answer should not automatically be treated as a correct one. Important information needs to be verified against official sources, original documents or trustworthy media.

Deepfakes and the manipulation of reality

Another challenge is artificially created content. Today's tools can generate photographs, video and audio that convincingly imitate real people or events. These materials are known as deepfakes.

The technology can be used for creative, educational and entertainment purposes. But it can equally be misused for fraud, disinformation, reputational damage or influencing public opinion.

A video that looks authentic is no longer necessarily proof that an event took place. Content has to be assessed by checking the original source, the date, the context, and other reporting on the same event.

The biases hidden in algorithms

AI systems learn from data created by people. If that data contains bias, gaps in representation or inequality, the system can reproduce those in its results.

This becomes especially important when AI is used for decisions that directly affect people's lives — employment, education, lending or public services.

Important decisions should therefore never be left entirely to an algorithm. What is needed is human oversight, transparency, and the ability to challenge a decision reached with the help of technology.

The role of journalism in the age of AI

For the media, artificial intelligence brings both new opportunities and new responsibilities. It can help journalists analyse documents, organise data, transcribe interviews and identify topics of public interest.

But speed cannot substitute for accuracy. Any material produced with the help of AI must be checked by a person before publication. Sources need to be verified, quotations compared against the original material, and the public informed when content has been created or substantially altered by artificial intelligence.

In an environment where anyone can produce convincing content, professional journalism has an even greater role: distinguishing verified information from manipulation.

How can we spot questionable content?

No single method guarantees detecting every piece of AI-generated material. Still, a few steps help:

• Check who published the content

• Look for the original source and context

• Compare the information against several trustworthy sources

• Do not rely on the headline, photo or video alone

• Pay attention to the date and place of publication

• Be wary of content designed to trigger an immediate emotional reaction

• Do not share information you have not verified

Critical thinking does not mean doubting everything. It means asking for evidence before accepting or passing on a claim.

Trust, but verify

Artificial intelligence should be seen neither as an absolute source of truth nor as a technology to be rejected outright. It is a powerful tool whose value depends entirely on how it is used.

In the future, the most important skill may not be using AI, but evaluating what it produces. Society will need citizens who know how to check sources, recognise manipulation, and demand transparency from the institutions and companies that use algorithms.

In the age of artificial intelligence, trust should not be automatic. It has to be built through accuracy, transparency and verification.