Meta CEO Mark Zuckerberg and OpenAI’s Sam Altman have been aggressively promoting the idea that everyone – children included – should form relationships with AI “friends” or “companions”. Meanwhile, multinational tech companies are pushing the concept of “AI agents” designed to assist us in our personal and professional lives, handle routine tasks and guide decision-making.

But the reality is that AI systems are not, and never will be, friends, companions, or agents. They are, and will always remain, machines. We should be honest about that and push back against misleading marketing that suggests otherwise.

The most deceptive term of all is “artificial intelligence”. These systems are not truly intelligent, and what we call “AI” today is simply a set of technical tools designed to mimic certain cognitive functions. They are not capable of true comprehension and are neither objective, fair nor neutral.

Nor are they becoming any smarter. AI systems rely on data to function, and increasingly, that includes data generated by tools like ChatGPT. The result is a feedback loop that recycles output without producing deeper understanding.

More fundamentally, intelligence is not just about solving tasks; it’s also about how those tasks are approached and performed. Despite their technical capabilities, AI models remain limited to specific domains, such as processing large datasets, performing logical deductions, and making calculations.

When it comes to social intelligence, however, machines can only simulate emotions, interactions, and relationships. A medical robot, for example, could be programmed to cry when a patient cries, yet no one would argue that it feels genuine sadness. The same robot could just as easily be programmed to slap the patient, and it would carry out that command with equal precision – and with the same lack of authenticity and self-awareness. The machine doesn’t “care”; it simply follows instructions. And no matter how advanced such systems become, that is not going to change.

Simply put, machines lack moral agency. Their behaviour is governed by patterns and rules created by people, whereas human morality is rooted in autonomy – the capacity to recognize ethical norms and behave accordingly. By contrast, AI systems are designed for functionality and optimization. They may adapt through self-learning, but the rules they generate have no inherent ethical meaning.

Consider self-driving cars. To get from point A to point B as quickly as possible, a self-driving vehicle might develop rules to optimize travel time. If running over pedestrians would help achieve that goal, the car might do so, unless instructed not to, because it cannot understand the moral implications of harming people.

This is partly because machines are incapable of grasping the principle of generalizability – the idea that an action is ethical only if it can be justified as a universal rule. Moral judgment depends on the ability to provide a plausible rationale that others can reasonably accept. These are what we often refer to as “good reasons.” Unlike machines, humans are able to engage in generalizable moral reasoning and can therefore judge whether their actions are right or wrong.

The term “data-based systems” ( DS ) is thus more appropriate than “artificial intelligence”, as it reflects what AI can actually do: generate, collect, process and evaluate data to make observations and predictions. It also clarifies the strengths and limitations of today’s emerging technologies.

At their core, these are systems that use highly sophisticated mathematical processes to analyze vast amounts of data – nothing more. Humans may interact with them, but communication is entirely one-way. DS have no awareness of what they are “doing” or of anything happening around them.

This is not to suggest that DS cannot benefit humanity or the planet. On the contrary, we can and should rely on them in domains where their capabilities exceed our own. But we must also actively manage and mitigate the ethical risks they present. Developing human-rights-based DS and establishing an international DS agency at the United Nations would be important first steps in that direction.

Over the past two decades, Big Tech firms have isolated us and fractured our societies through social media – more accurately described as “anti-social media”, given its addictive and corrosive nature. Now, those same companies are promoting a radical new vision: replacing human connection with AI “friends” and “companions”.

At the same time, these companies continue to ignore the so-called “black box problem”: the untraceability, unpredictability and lack of transparency in the algorithmic processes behind automated evaluations, predictions and decisions. This opacity, combined with the high likelihood of biased and discriminatory algorithms, inevitably results in biased and discriminatory outcomes.

The risks posed by DS are not theoretical. These systems already shape our private and professional lives in increasingly harmful ways, manipulating us economically and politically, yet tech CEOs urge us to let DS tools guide our decisions. To protect our freedom and dignity, as well as the freedom and dignity of future generations, we must not allow machines to masquerade as what they are not: us.

Peter G. Kirchschläger is a professor of ethics and the director of the Institute of Social Ethics ISE at the University of Lucerne and a visiting professor at ETH Zurich.

Copyright: Project Syndicate