Shipping executives say demand for low‑carbon marine fuels has softened this year as uncertainty over the International Maritime Organization’s (IMO) emissions framework persists and some firms opt to use carbon credits. IMO member states are due to meet before early November to discuss formally adopting a net‑zero framework to guide the industry to zero carbon by 2050. In Singapore, the world’s largest bunkering hub, government data show Jan–Sep sales of marine biofuels totaled about 603,000 to

2026-10-16

Shipping executives say demand for low‑carbon marine fuels has softened this year as uncertainty over the International Maritime Organization’s (IMO) emissions framework persists and some firms opt to use carbon credits. IMO member states are due to meet before early November to discuss formally adopting a net‑zero framework to guide the industry to zero carbon by 2050. In Singapore, the world’s largest bunkering hub, government data show Jan–Sep sales of marine biofuels totaled about 603,000 tonnes, down 47% versus the same period in 2025. Jens Maul Jorgensen, a director at dry‑bulk owner Oldendorff Shipping, said: “I do not expect biofuel demand to grow significantly before the IMO makes its decision.”

其他消息
2026-10-16

AI模型评测平台Arena负责人Peter Gostev于北京时间10月16日发布国际象棋实验复盘,指出当前大模型仅靠反复练习、保存记忆和延长上下文,仍难以实现稳定的持续学习。 实验先让GPT-6 Astra和Claude Opus 5.5分别与Stockfish棋类引擎对弈200局,自主选择难度、记录经验并调整策略。Opus最终获得约1760的估算Elo评分,高于Astra的1400(Elo评分是一种衡量棋手相对竞技水平的评分体系,广泛应用于国际象棋等竞技项目)。 但在随后的30局直接对抗中,Astra反而以24.5比5.5大胜。另一项对照测试显示,GPT-6.1 Sol使用约80万Token长上下文时,表现反而逊于约30万Token版本。 Gostev认为,模型在长时间运行中可能形成固定策略或反复改变战术,却未必能有效积累经验。实验反映出,增加上下文容量和交互次数,并不必然转化为持续学习能力。

2026-10-16

IMF Managing Director Georgieva warned that without safeguards the risks of AI could outweigh its opportunities.