AI, the fire of the 21st century: learning how to hold it without getting burned

Long before writing, our distant ancestors learned to master fire. It warmed them, fed them, protected them. It also burned them. This is the image Ariadna Font chose to describe artificial intelligence during the twelfth edition of the AI Congress, organized in Barcelona by the Eurecat technology center. "We must learn to control and govern it," she stated, according to the business media outlet VIA Empresa, before warning: fire is slipping out of our hands.

Her voice carries weight. Ariadna Font led engineering at IBM Watson, then headed responsible AI at Twitter, before founding the start-up Alinia AI. She is not a voice hostile to technology. She is someone who builds it from the inside, and who asks us not to play with matches.

The observation: a fire spreading to everyone

The term "artificial intelligence" was established in 1956, during a research workshop at Dartmouth College in the United States. For decades, the field advanced slowly, with periods of disinterest that researchers call "AI winters." Then, everything accelerated.

In 2016, Google DeepMind's AlphaGo program beat one of the world's best Go players. At the end of 2022, ChatGPT made generative AI accessible to everyone. Generative AI is AI that produces new content (text, images, sound) based on a simple written request.

Today, fire is everywhere. According to Stanford University's AI Index Report 2026, 88% of organizations reported using AI in 2025, compared to 55% in 2023. Global investments in AI reached approximately $580 billion in 2025, with about $345 billion in private investment.

Joan Mas, director of CIDAI, the Catalan center dedicated to data and AI, recalled during the same congress that this race is primarily between the United States and China. Europe, for its part, is adjusting its rules. The so-called "AI Omnibus" regulation entered into force on July 27, 2026. It simplifies the European AI Act and pushes back certain obligations: those targeting "high-risk" AI systems, for example in recruitment or education, will apply starting in December 2027.

The alert: the machine makes mistakes, you answer

Ariadna Font does not deny the power of the tool. She even believes, according to VIA Empresa, that AI is and will be more intelligent than us in many dimensions. But she adds a phrase that every leader should keep in mind: AI can multiply a team's productivity, but also the scope of an error.

She cites four concrete risks for a company:

• Data leakage: what you entrust to an AI tool can leave your company.
• Hallucination: the AI invents a false answer, but presents it with the assurance of truth.
• Agent actions: an AI agent is an AI entrusted with a mission that acts on its own (sending an email, placing an order). A wrong action can be costly.
• Content manipulation: faked texts, images, or voices.

Above all, she warns that it is useless to shift the blame to the machine: if a chatbot makes a mistake, the responsibility is yours. This is not an opinion; it is already jurisprudence. In February 2024, a tribunal in British Columbia, Canada, ruled against Air Canada. Its website chatbot had promised a grieving customer a reduced fare to which they were not entitled. The company argued that the chatbot was responsible for its own words. The tribunal rejected this argument: the company is accountable for all information published on its site, whether it comes from a web page or a conversational robot. The amount was modest, about 650 Canadian dollars. The principle, however, is immense.

This is where the question becomes ethical. Delegating a task to a machine is not delegating your conscience. You can entrust it with calculation, writing, and research. You cannot entrust it with responsibility.

The hope: taming the fire rather than extinguishing it

Humanity did not give up on fire because it burned. It invented the hearth, the fireplace, the lantern, the fire brigade, and safety rules. It learned to handle it. For AI, the path is the same.

Ariadna Font draws a positive conclusion: use this moment to promote all that it means to be human. The leaders of tomorrow, she says, must combine discernment and human values with digital fluency. And she warns against rushing: it is easy to get lost when you want to go too fast.

This idea aligns with that of the encyclical Magnifica humanitas, published in 2026, in which Pope Leo XIV addresses, among other things, artificial intelligence. Its central theme: building the city rather than the Tower of Babel. Moving forward, yes, but to serve humanity, and while keeping the reins.

And you?

In your company or your life, who reviews what the AI produces before it is sent to a client, a colleague, or a loved one? And if no one does, who will bear the responsibility the day it makes a mistake?

If this article was useful to you, subscribe to the allelu.ai newsletter: every Thursday evening, AI news explained simply, from an ethical perspective. And forward this article to someone who should read it.

This tension between an exponentially increasing power and our capacity to master it is at the heart of my book L'abondance tuera, available only on allelu.ai.

Sources

• VIA Empresa, "La IA, el fuego del siglo XXI que 'se nos escapa de las manos'", October 20, 2026: read the article
• Stanford HAI, AI Index Report 2026, Economy chapter: consult the report
• Dartmouth College, on the origin of the term "artificial intelligence": read the page
• Hunton Andrews Kurth, "EU Digital Omnibus on AI enters into force": read the article
• American Bar Association, Business Law Today, on the decision Moffatt v. Air Canada, February 2024: read the article
• Vatican News, on the encyclical Magnifica humanitas: read the article

Written with the help of AI, reviewed and validated by Reynald Naulleau.