Autopilot and AI: Misleading Metaphors
The Illusion of Simplistic Metaphors
Rapid advances in artificial intelligence, agentic AI in particular, often prompt the use of metaphors in an attempt to explain them. A common comparison is that of AI to airplane autopilot systems. While this may seem a convenient way to understand how these technologies can be integrated into daily life, the analogy is fundamentally flawed and misleading. Few metaphors manage to capture the real complexity of AI, and even fewer that of agentic AI, which often leaves the public with a distorted understanding of what is really going on behind the scenes.
In the near future, it could be tempting to use AI in the same way pilots use autopilot systems. However, several reasons make this comparison inappropriate:
-
First, the use of popular terms and technological complexity can give certain technologies an almost magical quality.
-
Second, most people do not really understand how autopilot works.
-
Finally, few people grasp the difference between simple systems and complex systems.
The Personal Use of AI and Its Implications
Before going further, it is important to clarify what is not being said. AI tools have their place in daily work: they make it possible to optimize and speed up certain aspects of a workflow, much like Figma, Excel formulas, a spell checker or StackOverflow. This use nevertheless calls for caution. AI can improve human lives, but only if it is democratized and the benefits are socialized for all. The ethical questions about how AI extracts and plagiarizes the work of others, or about the energy demands of these technologies, are not the main subject here. What is worrying is the way the technology is presented to the public and how this could help accelerate technofeudalism, to borrow Yanis Varoufakis's term. One can use products from Google, TikTok and Meta despite distrusting how they handle data privacy, and accept the implicit social contract, while arguing for a change in this balance of power. These technologies are revolutionary, and business practices and social structures must evolve to protect the public.
The Power of Words
Skepticism is warranted toward those who begin a statement with "In the future, we will..." when talking about new technologies. These predictions are rarely accurate, and believing that one alone holds premonitions about the technological future is vanity. The need to use metaphors to help others grasp a complex subject is understandable, but this first requires a deep understanding of the technology in question and of where the metaphor comes from. Have you ever thought about how many maritime expressions are used in corporate speak?
-
"You just have to learn the ropes."
-
"She runs a tight ship."
-
"This team is well-balanced."
-
"Let’s do a quick patch-up."
-
"Make sure everything is shipshape."
It may not be necessary to understand navigation to grasp these terms, but that is mainly because they are no longer tied to their original meanings. Having run a 16th-century ship is not a prerequisite for understanding these common terms. The term "autopilot" has met a similar fate; almost everyone is familiar with the concept (even children), although very few people hold a pilot's license.
At first glance, this parallel might seem to endorse "autopilot" as an acceptable metaphor, but there is an important distinction here. When someone "learns the ropes," the consequences of handling a ship's sails are not at stake; it is simply an expression. If you "run a tight ship" at home, the term does not directly imply any degree of navigation in your household. These terms are used without any inherent power dynamic. One could also say that these phrases have become so detached from their original meaning that they no longer relate to their original context. This may seem like nitpicking over semantics, but in the case of AI and how it is described or defined, the language used by executives and decision-makers represents a power imbalance. One can turn to the work of Noam Chomsky to understand this better; having the power to name something means controlling the narrative.
Le brief IA que les pros lisent chaque soir
Les 7 actus IA du jour, décryptées en 5 min. Gratuit.
Inclus dès l'inscription : notre sélection des meilleurs guides & comparatifs IA.
Choisissez votre rythme
Gratuit · Pas de spam · Désabonnement en 1 clic
Just consider a few more terms to see the difference and the spectrum. Euphemisms have the power to cloak their subject in ambiguity. Famously, the term "carbon footprint" was created by the gas and oil industry. Nazi Germany deployed the euphemism "Final Solution" to commit its atrocities and convince its citizens to comply. Many authoritarian governments have used "re-education camps" to imprison and torture dissidents. "Layoffs" or "readjustments" dehumanize the way some companies treat their staff and normalize it as part of the economic cycle. A century ago, "rights" meant that a citizen was entitled to certain rights and benefits; today, the term has been conflated with "being demanding," which carries a negative connotation and is used against those who rely on the welfare state. "The Cold War" was a useful term to justify the proliferation of nuclear weapons, American hegemony and the military-industrial complex. There are countless other examples.
Another pressing example is the rise of "bet on anything" markets. It is no secret that the legalization of sports betting has led to an explosion of gambling addiction, particularly among young men. The rise of Kalshi and other apps allows users to bet on almost anything, such as which song Bad Bunny would play first at the Super Bowl or whether the U.S. will invade Iran or Venezuela. Kalshi and other apps circumvent regulations by calling their services "prediction markets" rather than "gambling." Casinos and even sports betting are subject to far more oversight because of this semantic difference in naming. This has led to insider scams creeping into Kalshi. If you want to learn more, see Alex Falcone's work on sports betting and prediction markets. This simple linguistic distinction has allowed Kalshi to become a multibillion-dollar company almost overnight.
Artificial intelligence also benefits from a few etymological tricks. The term "Artificial Intelligence" anthropomorphizes the technology. Humans anthropomorphize anything that can reproduce speech, simply because speech is understood to make the human species unique. Yet tools like ChatGPT are essentially very sophisticated parrots. Moreover, the name suggests a deep, underlying "intelligence" that is rooted in science fiction. Similarly, as TikToker Adam Aleksic, also known as Etymology Nerd, points out, AI tools deploy "sparks" in their iconographic communication. This, combined with the use of the color "purple," has succeeded in presenting AI as "magic" to the general public rather than simply as extrapolated statistics and mathematics. Arthur C. Clarke noticed this in the 1960s when he said: "Any sufficiently advanced technology is indistinguishable from magic." While AI is not a scam like Kalshi or sports betting, its perception on the market is distorted by this language, which shapes and defines how this technology is understood.
In summary, language is a vector of communication, and the semantic meaning, etymology and epistemology associated with these words and phrases can provide cover for ideas and concepts that might otherwise be dangerous or harmful.
Demystifying Autopilot
It is safe to say that most people do not really understand autopilot technology. Many therefore subscribe to Clarke's Third Law, essentially believing that the plane flies itself and that pilots are just there to keep an eye on things and switch the seatbelt sign on and off. In reality, autopilot is a complex decision-making machine that provides pilots with feedback and information for the joint management of processes with ground control and other flights. As Tech Insider notes, the autopilot system is a negative feedback loop that seeks balance. Sensors distributed throughout the aircraft take continuous measurements and send them back to a central computer. This data is processed and fed back into the system to seek stability across multiple axes (vertical, horizontal, etc.).
Earl Wiener, a former pilot and aviation academic, coined the "Two D's of Automation." Autopilot is "Stupid and Diligent," meaning it will accept any type of acceptable input (even if it is illogical) and will always pursue its primary goal (to fly, land, etc.). In fact, this has become known as Wiener's Law. This means that a pilot must understand all available inputs and readings and constantly check for irregular results or behaviors. Understanding acceptable altitude levels, inclinations and angles, speeds, wind directions and much more allows the pilot to act somewhat like the "boss" of the autopilot. As Trevor Aldridge, a professional pilot, points out, autopilot frees a pilot to carry out higher-priority activities.
Brief IA — L'actualité IA en français
L'essentiel de l'actualité de l'intelligence artificielle, décrypté et expliqué chaque jour.