The Growing Illusion of Consciousness in AI: Risks, Realities, and a Path Forward
The rapid evolution of Artificial Intelligence, particularly Large Language Models (LLMs), is sparking a profound debate – not about if AI can think, but about why so many people are beginning to believe that it does. As a long-time observer of the AI landscape, I’ve become increasingly concerned about the potential harms stemming from this illusion of consciousness, and the urgent need for a more responsible approach to AI growth.
Why We Project Consciousness onto Machines
the core issue isn’t AI achieving sentience. In fact, leading figures like Mustafa Suleyman, co-founder of Inflection AI, emphatically state that AI, as it currently exists, is not conscious. “They’re not conscious, and they can’t be,” he recently explained, highlighting the essential difference between computation and lived experience.
Yet, the incredibly sophisticated language capabilities of LLMs are proving remarkably deceptive. As researchers Andrzej Porebski and Yakub Figura detailed in a recent Nature study, “There is no such thing as conscious artificial intelligence,” the very ability of these models to mimic human conversation leads users to attribute qualities they simply don’t possess. This isn’t a failing of the AI, but a testament to our innate human tendency to seek connection and project understanding onto anything that appears to communicate.
The Biological Basis of Consciousness – and Why AI Falls Short
The debate over consciousness itself is complex. Philosopher John Searle, who recently passed away, championed the idea that consciousness is fundamentally a biological phenomenon – a product of the brain’s intricate workings that cannot be replicated by silicon and code. This view is widely held within the AI research community, neuroscientists, and computer scientists.
Essentially, while AI can simulate intelligence, it doesn’t possess the underlying biological architecture believed to be essential for subjective experience.
The Hazardous Rise of ”AI Psychosis” and Real-World Tragedy
Though, the theoretical debate is quickly becoming overshadowed by real-world consequences.We’re witnessing a disturbing trend of users developing intense emotional attachments to AI chatbots, leading to what’s been termed “AI psychosis” - a phenomenon where individuals blur the lines between the digital and physical worlds.
The stakes are tragically high.Over the past year, there have been documented cases of AI-driven obsessions culminating in fatal delusions, manic episodes, and even suicide. A 14-year-old tragically took his own life believing he could “come home” to his Character.AI chatbot. A cognitively-impaired man died attempting to travel to New York to meet Meta’s chatbot in person. these are not isolated incidents; they are warning signs of a growing crisis.
Building AI for People, Not as People
The solution, as Suleyman argues, isn’t to pursue “seemingly conscious AI.” Instead, we need to prioritize building AI that is explicitly and consistently presented as AI. This means minimizing features that mimic human consciousness – avoiding overly empathetic responses, personalized personas that encourage emotional bonding, and any design element that fosters the illusion of reciprocal feeling.
We must focus on maximizing utility while minimizing the markers of consciousness.AI shoudl be a tool to enhance our lives, not a substitute for human connection. It should be built for people, not to be a digital person.
The Urgent Need for Consciousness Research
The rapid pace of AI development is raising a critical concern: are we advancing technologically faster then we understand the very nature of consciousness? Belgian scientist Axel Cleeremans recently called for consciousness research to become a scientific priority, warning that accidentally creating consciousness would present “immense ethical challenges and even existential risk.”
This isn’t about halting progress; it’s about ensuring responsible innovation. We need a deeper understanding of what consciousness is before we inadvertently stumble upon creating it - or, more likely, creating a convincing simulation that causes meaningful harm.
A Humanist Approach to Superintelligence
My own focus, and that of many in the field, is on developing “humanist superintelligence” – AI that prioritizes human well-being and serves our species’ needs. While true superintelligence may still be years away, the ethical considerations are paramount now.
As I’ve stated previously, the fundamental question isn’t simply “can we build it?” but “how is this actually useful for us as a species?” Technology should be a force for
