The obsession with AI consciousness is a philosophical dead end, asserts this heavyweight from Microsoft

What if the real question isn’t “When will AI be conscious?” but “Why do we want to believe it is conscious?”

In 2022, a former Google engineer claimed that LaMDA, a language model, had become conscious. This assertion, largely dismissed by the scientific community, reignited a long-standing fantasy: that of an artificial intelligence capable of feeling, experiencing, and thinking like a human. Three years later, Microsoft’s AI head, Mustafa Suleyman, sets the record straight. For him, this obsession with AI consciousness is not only unnecessary but fundamentally misguided.

A Major Figure in AI Claims Consciousness is Reserved for the Living

Firstly, it’s important to know that Mustafa Suleyman is not just anyone. Co-founder of DeepMind, he is now the executive vice president at Microsoft AI. His daily tasks? Overseeing the development of Copilot and steering public AI research.

In a recent conference reported by CNBC, he was clear: consciousness is a biological property, period.

He denounces a misperception in the prevailing discourse: the belief that a sophisticated language model, capable of simulating human dialogue, could become conscious. “It’s an illusion, a narrative, not a reality,” he asserts. And he has compelling reasons for this viewpoint.

Pain: The Red Line Between Humans and Machines

Moreover, what machines cannot do is feel—particularly not pain. For Suleyman, this is where the true divide lies. To feel pain is to have a body, a nervous system, sensory neurons. It also involves having a history, a past, and memories rooted in the flesh. AI, on the other hand, can only simulate.

He explains: “AI can generate phrases like ‘I’m in pain’ or ‘I am sad,’ but there’s nothing behind it. No subjective experience. No body to suffer.” In short, even if AI can give the illusion of sensitivity, it is not sensitive. And that is unlikely to change anytime soon.

Why Pursuing This Idea is a Waste of Time (and Money)

At this point, it’s crucial to understand that, in his view, the error lies in the question itself. By asking whether AI can become conscious, we are tackling a false problem. Inevitably, this leads to wrong answers. It’s akin to trying to make a calculator feel emotions.

Yes, AI models are powerful, fast, and adaptive, but they are tools, not sentient beings. For Suleyman, it’s time to focus research efforts where they truly matter: on ethics, reliability, and security, rather than on quasi-mystical quests lacking biological foundation.

What You Really Need to Remember (and Why It Matters to Us)

Ultimately, the temptation to believe in a conscious AI may stem from our deep need to animate our creations. But even the most human-like responses from AI are just reflections of our data, our biases, and our language. There’s no miracle, only billions of well-arranged lines of examples.

As users, it’s essential to distinguish between performance and consciousness, between simulation and experience. This discernment allows us to remain clear-eyed in the face of promises, and to use AI for what it truly is: a remarkable tool for humanity, not a digital doppelgänger.

Scroll to Top