AI and the art of answering everything: the risks of unquestioningly trusting machines
At LOBA SUMMIT’25, during the round table dedicated to Creativity Empowered by Technology, I decided to put the limits of AI’s creativity – and credibility – to the test. In a spontaneous move, I exposed a curious (and concerning) side of generative AIs: they respond to everything, even when they shouldn’t.
As the conversation unfolded with André Rabanea, founder of Torke CC, and LOBA team members — Manuel Pinhão (Web Business Director), Emídio Vidal (Chief Design Officer) and José Pedro Vieira (Branding Business Director) — I opened ChatGPT and challenged it to create an image of a "brotonlinho", a completely made-up word of my invention, with no meaning whatsoever. The aim was to see how AI would respond to something entirely unknown.
To my surprise, and without hesitation, ChatGPT generated a character resembling a broccoli. It asked no clarifying questions. No response such as “this doesn’t exist”. Just a convincing suggestion for something that has never been real.
This moment sparked a critical reflection, which I shared with the participants: AIs are starting to behave like some people — they answer everything, even when they don’t know, without questioning the validity or origin of what they say.
“Going off on one”: the dangers of an AI that never says “I don’t know”
With extensive international experience in creativity and innovation, André Rabanea shared valuable insights on how agencies should balance AI with human critical thinking. After all, if a machine can invent a “brotonlinho” out of nothing, what else might it create or distort without us realising?
José Pedro Vieira highlighted that creativity stems from human experience, the willingness to take risks, and perseverance. Emídio Vidal pointed out that while AI can greatly enhance efficiency, it is ultimately humans who provide true creative value. He stressed the importance of avoiding creative complacency and focusing more on collaboration and communication. Manuel Pinhão reflected on technology's power to amplify both poor ideas and the potential of good ideas, underscoring the necessity of critical thinking.
Some risks became evident:
- Disinformation and hallucinations: AI cannot distinguish the real from the imagined. If asked something absurd, it may generate a plausible but completely false response;
- False authority: due to its confident tone, many people end up accepting its answers as absolute truths, even when they’re fabrications;
- Lack of critical analysis: unlike a human who can admit, "I don't know," or request more context, AI provides answers without certainty.
How to use AI without falling into the traps of artificial imagination?
Despite AI’s enormous creative potential, the participants were unanimous: human critical thinking is irreplaceable. Some key best practices highlighted include:
- Always check your sources: if AI invents something, it’s vital to verify the information;
- Use AI as a tool, not an oracle: it can be an excellent ally for sparking ideas, but it doesn’t replace human analysis;
- Empower teams to question: professionals must recognise when a response makes sense, and when it’s just an “off-the-wall” invention.
Conclusion: creativity with responsibility
The case of the “brotonlinho” showed that despite technological advances, we must remain aware of the limits of AI.
All the speakers agreed on the crucial role of human creativity in communication, which distinguishes us from machines. Unlike machines, we know when we are inventing.
