The AI Bubble - Between Technological Revolution and Financial Mirage
In the last year, the term "Artificial Intelligence" has become a kind of universal spice: it is added to any news story, any financial report and any product presentation to instantly increase its value. However, to a careful observer, the current frenzy begins to resemble strikingly moments in history that we would prefer not to repeat, such as the "dot-com" bubble of the late 1990s.
Until a few years ago, "a picture was worth a thousand words". Today, a picture can only be the result of a text command given to a software.
Lie: We see on social networks old "heroes", unreal landscapes or political events that never happened, all generated to stir emotions and collect likes (which are later monetized).
Impact: We lose trust in everything we see. When everything can be fabricated, nothing feels authentic, leading to a state of apathy and social confusion.
Let’s explore why we feel like we’re dealing with a “blatant lie” and where reality ends and aggressive marketing begins.
Everything is “AI,” even what it isn’t...
The biggest source of confusion is the massive rebranding. Basic algorithms that we’ve been using for decades (like spam filters or recommendation systems) have suddenly been rebranded “AI” to attract investors. Many startups that claim to use advanced AI are either running simple “if/then” algorithms or, in extreme cases, poorly paid people manually entering data (so-called “Indian-operated AI”).
The news tells us every day that AI will replace all jobs and create wealth overnight. But the reality on the ground is that implementing generative AI in companies is expensive, difficult to manage, and often prone to errors (hallucinations). Many companies invest huge sums only to find that the result is a chatbot that sometimes gives wrong information to customers.
There is a huge financial interest in keeping this bubble going for as long as possible.
Tech companies need a justification for their huge stock prices.
News publications need bombastic headlines to get clicks (“AI has discovered the secret to immortality!”).
Consultants need a complex new technology to explain to clients for large sums of money.
Is it all a lie? Not really - The technology behind large language models (LLMs) is fascinating and extremely useful in programming, translation, data synthesis, and scientific research. The lie is not the technology itself, but the “tertip” that it is a magical, cheap and infallible solution to any human problem.
History teaches us that a bubble bursts when massive investments do not translate into real profits. Currently, billions of dollars are spent on graphics chips (Nvidia) and electricity to train models, but the revenues generated directly by AI applications are still modest compared to the investment.
Like any bubble, this one will also clear: companies that do not bring real value will disappear, and AI will remain among us not as a deity that solves everything, but as a useful but limited tool, discreetly integrated into our daily lives.