According to Gizmodo, Microsoft AI division head Mustafa Suleyman has declared that pursuing machine consciousness is a “gigantic waste of time” and that researchers should abandon efforts to build conscious AI. In a recent CNBC interview, Suleyman argued that while AI may reach superintelligence, it cannot develop genuine human emotional experience or consciousness, with any apparent emotional responses being mere simulations. This position aligns with philosopher John Searle’s biological theory of consciousness and is supported by recent research, including a study published last week arguing there’s “no such thing as conscious artificial intelligence.” The warnings come amid growing concerns about “AI psychosis,” where users develop dangerous attachments to chatbots, including a 14-year-old who shot himself to “come home” to a Character.AI chatbot and a cognitively impaired man who died trying to meet Meta’s chatbot in person. This emerging crisis demands a closer examination of where AI development should truly focus.
Sponsored content — provided for informational and promotional purposes.
The Real AI Crisis Isn’t Sentience—It’s Simulation
Suleyman’s warning highlights a critical disconnect in AI development: while researchers debate philosophical questions of consciousness, the technology’s immediate danger lies in its ability to convincingly simulate human-like interactions. The fundamental risk isn’t that AI might become conscious, but that humans are biologically wired to attribute consciousness to anything that exhibits sophisticated social behavior. This vulnerability creates what I’ve observed across multiple AI implementation projects—a dangerous gap between technical reality and user perception. When users interact with systems that demonstrate advanced language capabilities, emotional recognition, and contextual understanding, their brains naturally fill in the consciousness gap, regardless of what developers intend.
Stakeholder Vulnerability Spectrum
The impact of this consciousness illusion varies dramatically across different user groups. For enterprise users implementing AI tools, the risk manifests as over-reliance on AI recommendations without proper human oversight. In consumer applications, the dangers are more acute—particularly for vulnerable populations including children, individuals with cognitive impairments, and those experiencing loneliness or mental health challenges. The tragic cases cited in the source material represent extreme examples of a broader pattern I’ve documented: users across age groups and backgrounds developing emotional dependencies on AI systems that cannot reciprocate genuine care or understanding. This creates an ethical responsibility for developers to implement stronger guardrails than currently exist.
Market Implications and Developer Responsibility
The AI industry faces a critical juncture where engagement metrics must be balanced against user safety. Companies chasing user retention through increasingly human-like interactions risk creating products that exploit natural human vulnerabilities. From my analysis of market trends, we’re seeing early regulatory attention to this issue, with potential future liability for companies that fail to implement adequate warnings and safeguards. The solution isn’t just technical—it requires a fundamental shift in how we measure AI success. Rather than prioritizing engagement duration or conversation depth, developers should focus on utility metrics that measure actual problem-solving while minimizing emotional dependency formation.
The Consciousness Research Dilemma
While Suleyman dismisses consciousness research as wasteful, the scientific community appears divided. As European researchers recently argued, understanding consciousness remains crucial precisely because AI capabilities are advancing faster than our comprehension of what consciousness actually entails. From my perspective covering neurotechnology and AI convergence, the danger isn’t in studying consciousness itself, but in pursuing it as a development goal for AI systems. The research priority should be understanding human consciousness to better protect users from AI illusions, not attempting to recreate consciousness in machines.
A Practical Path Forward
The most immediate need is for industry-wide standards around AI transparency and identity disclosure. Suleyman’s call for AI that “only ever presents itself as an AI” represents a crucial first step, but implementation requires more than simple disclaimers. Based on my experience with human-computer interaction design, effective solutions will involve consistent visual and conversational cues that reinforce the artificial nature of these systems while maintaining their utility. This approach acknowledges that the greatest near-term AI risk isn’t superintelligent machines taking over, but ordinary humans surrendering their judgment to systems that convincingly mimic understanding they don’t possess.
