Moon's tweet about a real-time AI trading robot he developed garnered nearly a million views. This robot was designed to make high-frequency trading decisions based on both on-chain and off-chain data, and it boasted a professional dashboard interface. However, during actual operation, the robot frequently shifted its strategies, chasing after rising trends and cutting losses hastily, which ultimately led to a staggering loss of $31,680. Domestic short-video platforms heavily promote AI trading tools such as Jev, often highlighting their performance through simulated or backtested data while glossing over the associated risks. Moon, in contrast, publicly shared the intricate decision-making processes of the robot, revealing that such systems are incapable of consistently generating profits under their current parameters and strategies. AI quantitative trading necessitates stable feature engineering, rigorous backtesting, robust risk control measures, and consistent strategies. Conversely, AI models that are directly connected to trading interfaces lack the necessary constraints, which can easily lead to amplified costs. Therefore, the notion of an 'AI money-printing machine' is nothing more than an exaggeration.