Can artificial intelligence truly be random? It's a question that delves into the heart of what it means to be human – and what it means to build truly intelligent machines. Humans, with our inherent biases and pattern-seeking tendencies, are notoriously bad at generating random sequences. We see patterns where there are none, and our choices, even when we strive for randomness, are often predictable. Now, a fascinating research paper, "How Random is Random? Evaluating the Randomness and Humaness of LLMs' Coin Flips," examines how Large Language Models (LLMs) fare in this arena. The results are surprising. Researchers tasked several LLMs, including GPT-3.5, GPT-4, and Llama 3, with the simple task of simulating coin flips. They analyzed the resulting sequences, looking for patterns and biases that deviated from true randomness. What they found was that the newer, more sophisticated models, like GPT-4 and Llama 3, actually amplified the human biases present in their training data. These LLMs showed an overwhelming preference for heads, especially on the first flip of a sequence, and tended to avoid long runs of either heads or tails. Interestingly, the older GPT-3.5 model exhibited more random behavior, suggesting that in the pursuit of human-like text generation, LLMs might be becoming *too* human. This raises a crucial question: is mimicking human bias a desirable trait in AI? In some contexts, like collaborative games or predicting human behavior, shared biases can be beneficial. But in other situations, such as generating secure passwords or conducting scientific simulations, true randomness is essential. The research highlights the importance of understanding how LLMs interact with randomness and the need for developing strategies to control their behavior. As AI systems become increasingly integrated into our lives, we must ensure they can be both human-like and truly random when the situation demands it. The ability to generate random sequences might seem like a small detail, but it's a fundamental test of an AI's capabilities – and a window into the complex relationship between humans and the machines we create.