This article originally appeared on Engadget at https://www.engadget.com/ai/big-tech-companies-agree-to-not-ruin-your-electric-bill-with-ai-data-centers-230102956.html?src=rss
As I referenced earlier, we don’t think digital games are just a flash in the pan market. We’re going to be patient, we’re going to invest, and if we lose money in the short term, we’re going to build up our community, build up our capacity, and build up our ability to be able to execute with excellence. And I think over time, it’s going to be a difference maker for us and something that we’re happy we invested in.
。关于这个话题，新收录的资料提供了深入分析
2024年，亏损逐季收窄：Q1亏损7.65亿、亏损率13.5%；Q2亏损收窄到6.08亿、亏损率9.9%；Q3净亏损收窄至7981万；Q4净利润8893亿、利润率 1.1%；。新收录的资料对此有专业解读
The BrokenMath benchmark (NeurIPS 2025 Math-AI Workshop) tested this in formal reasoning across 504 samples. Even GPT-5 produced sycophantic “proofs” of false theorems 29% of the time when the user implied the statement was true. The model generates a convincing but false proof because the user signaled that the conclusion should be positive. GPT-5 is not an early model. It’s also the least sycophantic in the BrokenMath table. The problem is structural to RLHF: preference data contains an agreement bias. Reward models learn to score agreeable outputs higher, and optimization widens the gap. Base models before RLHF were reported in one analysis to show no measurable sycophancy across tested sizes. Only after fine-tuning did sycophancy enter the chat. (literally)。关于这个话题，新收录的资料提供了深入分析