All Posts

10 posts

DeepSeek Beats GPT-4 in US Stock Trading—— HKU's AI-Trader Experiment Reveals True AI Investment Capabilities

·ai

University of Hong Kong's open-source AI-Trader project delivers shocking results—— DeepSeek achieved 9.68% returns in real US stock trading, outperforming GPT-5, Claude, Gemini and other top AI models. Five AIs received $10,000 each to autonomously trade NASDAQ 100 stocks for one month with zero human intervention. Deep dive into experimental design, strategy differences, why Gemini's high-frequency trading failed, and complete AI-Trader open-source features. Includes GitHub link and deployment guide.

DeepSeek美股交易收益9.68%碾压GPT：AI炒股实验完整解析

·ai

香港大学开源项目AI-Trader实验结果震撼——DeepSeek在真实美股交易中以9.68%收益率击败GPT-5、Claude、Gemini等顶级大模型。五个AI模型各获1万美元，在纳斯达克100市场自主交易一个月，零人工干预。本文深度解析实验设计、各模型策略差异、Gemini高频交易失败原因，以及AI-Trader开源项目的完整功能。附GitHub地址和部署指南。

AI Pioneer Geoffrey Hinton Warns —— Superintelligence, Job Loss, and Why We Must Act Now

·ai

Geoffrey Hinton, the Turing Award-winning "Godfather of AI," left Google in 2023 to warn humanity about AI's existential risks. This comprehensive analysis covers his full interview on The Diary Of A CEO, discussing superintelligence threats (10-20% extinction probability), cyber attacks, lethal autonomous weapons, job displacement, and why digital minds may surpass humans. Hinton explains why AI can't be stopped, why regulation must force companies to prioritize safety, and offers stark career advice:"Train to be a plumber." Essential reading for understanding AI's biggest dangers from the pioneer who helped create it.