I've noticed a recurring pattern in the way we talk about artificial intelligence. Companies like OpenAI often describe their latest models as if they "think" before answering, as seen with the recent release of OpenAI o1. This anthropomorphisation isn't just a harmless metaphor; it's a fundamental misrepresentation of what AI is and how it operates. And it's causing more harm than we might realise. At its core, AI doesn't think—**it processes and predicts**. These models analyze vast amounts of data and generate responses based on statistical patterns. When we say an AI "thinks," we're attributing human qualities to a machine that operates on fundamentally different principles. This isn't just a semantic quibble; it's a distortion that misleads both the public and professionals about the capabilities and limitations of AI. Consider the difference between **memorisation and understanding.** An AI can store and retrieve information, even solving problems at a PhD level in certain domains. But this isn't understanding in any meaningful sense. It's **pattern recognition and regurgitation.** Knowledge, in this context, isn't the same as intelligence. Intelligence involves comprehension, reasoning, and the ability to navigate novel situations—qualities that AI, as it currently stands, doesn't possess. The danger of anthropomorphising AI is that it sets unrealistic expectations and distracts from the real challenges and opportunities in the field. By pretending that AI can think like humans, we overlook the unique strengths of these systems as powerful tools for computation and analysis. More importantly, we risk neglecting the critical work needed to advance AI in meaningful ways—like improving algorithmic transparency, addressing biases, and ensuring ethical applications. There's also a kind of snake oil salesmanship at play here. Overhyping AI capabilities can be a marketing strategy, but it's one that ultimately undermines trust. When people realize that AI doesn't live up to the human-like attributes it's been ascribed, skepticism grows—not just about specific products but about the field as a whole. This skepticism can hinder investment, regulation, and public acceptance of genuinely beneficial technologies. We need to stop anthropomorphizing AI and start appreciating it for what it is. **AI is a tool**—a remarkably powerful one—that excels at processing data and identifying patterns beyond human capacity. By acknowledging its true nature, we can focus on leveraging its strengths and addressing its weaknesses without the confusion that comes from misleading metaphors. The real progress in AI won't come from trying to make machines think like humans but from enhancing their ability to assist us in ways that complement our own abilities. Let's reserve human qualities for humans and appreciate AI as the sophisticated, non-sentient technology it is. Only then can we have honest conversations about its development, manage expectations appropriately, and harness its potential responsibly. [[Essence of Intelligence - Prediction and Learning]] [[Essence of Mathematical Reasoning]] [[Architecture of LLM]]