意外

World Food Program appeals for $19 million to provide emergency food for Afghanistan_我的网站

一 |     ISLAMABAD -- The United Nations' World Food Program on Wednesday appealed for $19 million to provide emergency assistance to tens of thousands of people affected by a series of devastating earthquakes and aftershocks that has rocked western Afghanistan.Ana Maria Salhuana, deputy country director of the World Food Program in Afghanistan, said it was helping survivors but it urgently needed more funding because “we are having to take this food from an already severely underfunded program.”The group said it is working to provide emergency food assistance to 100,000 people in the region.“Disasters like these earthquakes pound communities who are already barely able to feed themselves back into utter destitution," the WFP said.A 6.3 magnitude earthquake struck part of western Afghanistan on Sunday, after thousands of people died and entire villages were flattened by major quakes a week earlier. It was the fourth quake the U.S. Geological Survey has measured at 6.3 magnitude in the same area in just over a week.The initial earthquakes on Oct. 7 flattened whole villages in Herat province and were among the most destructive quakes in the country’s recent history.The WFP said staffers responded within hours of the first earthquakes, distributing fortified biscuits, pulses and other food items to affected families in destroyed villages.“An estimated 25,000 buildings have been destroyed," the group said a statement. “The survivors are currently sleeping in tents next to the rubble of their homes, desperate and afraid of further earthquakes and aftershocks.”The latest quake was centered about 30 kilometers (19 miles) outside the city of Herat, the capital of Herat province, and was 6 kilometers (4 miles) below the surface, the U.S. Geological Survey said. More than 90% of the people killed were women and children, U.N. officials said. The quakes struck during the daytime, when many of the men in the region were working outdoors.Taliban officials said the earlier quakes killed more than 2,000 people across the province. The epicenter was in Zenda Jan district, where the majority of casualties and damage occurred.The WFP said affected families will need help for months with winter just weeks away. It said that if there is funding, the emergency response will be complemented by longer-term resilience programs so vulnerable communities are able to rebuild their livelihoods.The UN body was forced earlier this year to reduce the amount of food families receive and to cut 10 million people in Afghanistan from life-saving food assistance due to a massive funding shortfall. In addition to the earthquake response, the WFP also urgently needs $400 million to prepare food before winter, when communities are cut off due to snow and landslides. In Afghanistan, these include communities of women who are being increasingly pushed out of public life.The initial quake, numerous aftershocks and a third 6.3-magnitude quake on Wednesday flattened villages, destroying hundreds of mud-brick homes that could not withstand such force. Schools, health clinics and other village facilities also collapsed.Besides rubble and funerals after that devastation, there was little left of the villages in the region’s dusty hills. Survivors are struggling to come to terms with the loss of multiple family members and in many places, living residents are outnumbered by volunteers who came to search the debris and dig mass graves.Earthquakes are common in Afghanistan, where there are a number of fault lines and frequent movement among three nearby tectonic plates.。     8 月 24 日消息，据外媒 Phys.org 今天（24 日）报道，卡迪夫大学和墨尔本大学的一项新研究发现，生成式人工智能（GenAI）目前还无法像人类教师一样可靠地评判学生的书面作业。研究人员使用 ChatGPT 的两个版本，对 50 篇生物科学专业本科生论文进行评分，以测试大模型评估学生作业的能力。ChatGPT 需要按照 7 项标准批改这些论文，研究人员还采用了 4 种不同的提示方式，随后将大模型与人工评分的平均分和分数波动情况进行比较。

二 | 图源：Pexels卡迪夫大学生物科学学院威廉 · 凯博士指出：“研究结果显示，模型给出的分数波动很大，无法准确预测人工评分。生成式 AI 与人类批改同一批论文时存在明显差异。只看论文总分，两者给出的结果相对接近；一旦具体到每项评分标准和每一篇论文，大模型与人工评分之间的差距就相当明显。”除一种情况外，AI 模型给出的平均分普遍高于人工评分。平均分方面，AI 模型与人工评分的最大差距达到 16.1 分；具体到单篇论文，最大差距达到 40 分。获悉，威廉 · 凯还发现，AI 往往会压低高分论文的成绩，又把低分作业的成绩抬高，最终使分数系统性地向中间集中。研究结果说明，至少现阶段，即使经过大量训练，AI 仍无法可靠地对主观性较强的书面作业给出与人类相当的分数。

三 | “我们测试的大模型目前并不适合取代教师，为学生预测作业成绩。”研究人员指出，高等教育领域确实很关注大模型能否利用模式识别能力，让学生作业评分更加客观，同时提高批改效率、减轻教职人员压力。但这项研究表明，目前并不适合这么做。此外，未经学生明确同意便把作业提交给 AI 工具，本身还涉及伦理问题；大模型也不能、也不应该被依赖来给学生的长篇书面作业评分。“随着大模型继续发展，未来模仿人类判断的能力可能会提高。但从这项研究来看，要让 AI 给出的分数与人工评分真正保持一致，恐怕并不容易。

Current article:http://90h.cibianpangjujiusu.cyou/dhf2/fr8.html

Published on:11:14:17