机器之心报道机器之心编辑部一直以来，AI 领域的研究者都喜欢让模型去挑战那些人类热衷的经典游戏，以此来检验 AI 的「智能程度」。例如，在 Atari 游戏、围棋（如 AlphaGo）或《星际争霸》等环境中，游戏规则明确，边界清晰，研究者可以精确控制变量（如难度、初始状态、随机性等），确保实验的可重复性。而 AlphaGo ...

Waymo, America’s only fare-collecting robotaxi firm, runs over 700 cars in San Francisco, Phoenix and Los Angeles. Tesla, ...

The giant chipmaker has transformed its annual developer conference from an academic event into a who’s who gathering for the ...

The verdict is in: OpenAI's newest and most capable traditional AI model, GPT-4.5, is big, expensive, and slow, providing marginally better performance than GPT-4o at 30x the cost for input and ...

Forbes contributors publish independent expert analyses and insights. I write about entrepreneurship, AI, and the future of work.

However, GPT-4.5 is more aligned with the output from ... last week’s belabored release makes no sense; it’s super expensive and offers marginal gains only seasoned chatbot users may notice.

We explore the evolution of artificial intelligence, drawing parallels between Stanisław Lem's speculative fiction and modern ...

So far, GPT-4.5 has proven more accurate than GPT-4, with a Simple QA accuracy of 62.5% and a hallucination rate of 37.1%. It also exhibits improved naturalness and intent understanding and ...

The current version of ChatGPT is based on the GPT-4 model, which was trained on all sorts of written content including websites, books, social media, news articles, and more — all fine-tuned in ...