This is a guest post by Barbara Gabriella Renzi and Giulio Napolitano who have just published an article exploring AI metaphors from a bilingual perspective, comparing metaphors in English and Italian. This adds a new dimension to my own work on AI metaphors here on the Making Science Public blog and fills a real gap in research. To understand how metaphors for AI function in society we need more work on metaphors in and across different languages.
•••
Thinking about AI through metaphors
Artificial intelligence is definitely in the air, especially in the form of generative AI language and image models: they can create articles and paintings, forecast the weather and recommend movies and music. They can even apologise! However, despite the growing capabilities and visibility of these systems, their underlying mechanisms and processes remain largely unknown to the public. Perhaps this is why we often talk about AI using metaphors.
We call AI a brain, a black box, a mirror, an assistant or even a modern oracle. And these metaphors are more than simply “colourful language”; they correspond to ways of thinking. Ways of thinking about something that is difficult to visualise or understand. Metaphors are valuable in translating unfamiliar technologies into familiar experiences and, in doing so, they shape how we understand AI.
In search of AI metaphors
We collected some metaphors referring to generative AI that appear in English and Italian in online sources: newspapers, blogs and forums. Our goal was not to provide a definition of what AI is, but to see how AI is described in a sample of everyday communication. The results were intriguing.
Some metaphors were clustered around a few general themes. Others included depictions of AI as a thinking agent, a creative tool for human minds, a revolution, as well as mythological or philosophical themes. In other words, AI seems to be understood through multiple metaphorical lenses.
AI as a thinking system
A familiar group of metaphors compares generative AI to human cognition: the brain, the thinking machine, the artificial eye and the well-known black box. In our vocabulary, words like “complexity”, “bias”, and “black box” have all started to crop up in everyday conversation. This has been especially true of the black box metaphor. While many intuitively feel that they can follow and understand the inputs and outputs of a system that behaves like a human in some respects, at the same time willing to accept that anything in between remains a mystery, this “in between” has become the heart of innumerable discussions around the topic of transparency and explainability in AI and machine learning.
This metaphor theme also came up in the Italian material. For instance, it explains that an autopilot is a system that flies a plane, but a human must be present in the cockpit to monitor it. So, in this case, the metaphor portrays AI as a tool that augments the decision-making capabilities of humans. Both metaphors address questions of trust, but in slightly different ways.
AI as a creative partner
Another related cluster of metaphors presents AI as an enhancement of human capabilities. There, the AI stands out as a digital consultant, a virtual assistant or a creative brush. Maybe this not unexpected, with generative AI increasingly used in many domains for writing, designing and learning. In this framing, AI is not necessarily competing with human intelligence but instead collaborating with it. The emphasis shifts from automation to augmentation.
AI as transformation
Some metaphors are much grander. Some call the new technology at the heart of AI the “fuel” of a new industrial revolution, others see it as a “catalyst” that triggers changes, and others call it a “tectonic force” that reshapes ecosystems. These metaphors position AI in a developmental context, bearing similarities to the metaphors used in relation to electricity, nuclear power or the internet. But they also raise questions about disruption and control.
AI as reflection
The mirror-reflecting-AI metaphor has popped up a lot as well. In this framework, AI is seen as a reflection of the patterns existing in the training data. If there is a bias in an AI system, it is then seen as a mere reflection of biased human behaviour and societal structures. The mirror metaphor reframes the discussion from being about the technology that employs the data to being about the society that produces the data. It invites a moment of introspection.
AI as a modern oracle
Italian sources also produced some more ancient and mythological imagery. One example was the metaphor of AI as a modern oracle. In ancient times, oracles were said to reveal deep mysteries and foretell the future, although their answers were often cryptic and open to interpretation. There’s definitely a parallel here with today’s AI systems that can predict trends and behaviours but are almost as impenetrable. The black box acts as a type of oracle, providing responses without explaining how those conclusions were drawn. In the end, this kind of metaphor evokes both fascination and unease.
Different cultural lenses?
Finally, when comparing the English and Italian material, an interesting contrast emerged. In the world of English, there were many metaphors related to the practical application of Artificial Intelligence: AI as a “tool”, “assistant” or “productivity engine”. Instead, in Italy there were metaphors derived from philosophy, mythology and culture. While English metaphors tend to focus on what AI does, in Italian they tend more to explore the question of what AI is.
If you are intrigued, you can find our full research here:
Renzi, B.G., Napolitano, G. (2026). AI Through a Bilingual Lens: Unveiling Metaphors in Italian and English AI Discourses. In: Karamifar, B., Valente, A.C. (eds) Artificial Intelligence and Discourse. Postdisciplinary Studies in Discourse. Palgrave Macmillan, Cham. https://doi.org/10.1007/978-3-032-06408-0_4
Image: Pexels. “Foto Cermin Persegi Panjang” (Rectangular Mirror Photo) uploaded by photographer Darya Sannikova
Making Science Public 
Leave a comment