#预测市场 Seeing Gemini's win rate on Kalshi soar from 30% at the beginning of the year to 86%, I have to say a few honest words. This is not good news; rather, it serves as a warning — prediction markets are inherently susceptible to emotions and capital flows.

At the start of the year, ChatGPT had a 41% probability, now it has dropped to 8%, and Grok has fallen from 14% to 6%. These numerical changes seem scientific, but in reality, they are driven by capital chasing hot topics. Think about it — the recent surge in public opinion around Gemini 2.0 naturally attracted people to bet, pushing the probability higher. This is the same logic as the "consensus boosting" in the crypto circle.

Let me give a reminder — don’t treat prediction markets as the absolute truth. $14 million in trading volume sounds substantial, but that money is either from institutions arbitraging or retail gamblers. There’s still over a year until settlement; who dares to say that the current lead will last until the end of 2025? The pace of change in AI is so fast that you can figure it out with your toes.

The most dangerous thing is when people look at this 86% figure and decide to bet, thinking "the probability is so high, it’s a sure thing." Little do they know, prediction markets are just emotion amplifiers. When big funds come in to dump, retail investors are the ones getting chopped up like leeks. Staying vigilant is far more important than blindly following the trend.