作者：百年不渡分类：未来幻想点击：5852699次下载：5826294次大小：70M 日期：2026-08-26

白夜追凶

卡迪夫大学新研究：AI 尚无法像人类教师一样可靠地批改作业_我的网站

五星酒店潜规则

一 | Islamabad, Oct 21 (UNI) In a major legal victory for the ruling coalition on Friday, the Election Commission of Pakistan (ECP), in a consensus verdict in Toshakana reference, disqualified former prime minister Imran Khan and ruled that the PTI chairman is no more a member of National Assembly.

Criminal proceedings will be initiated against the PTI chairman for misdeclaration, the verdict said.

The ECP stated that Imran Khan submitted a false affidavit and was found involved in corrupt practices under Article 63(1)(p).

Under the said article, a lawmaker is disqualified for the time being from being elected or chosen as a member of the Majlis-e-Shoora (parliament) or of a provincial assembly.

The decision was taken unanimously by the ECP’s five-member bench but the Punjab member Babar Hassan Bharwana was not present today as he was unwell.

"As sequel to our abovementioned findings, facts available on record and keeping in view the argument of learned counsel for parties herein, we are of the considered opinion that the Respondent has become disqualified under Article 63(1)(p) of the Constitution read with Section 137,167 and 173 of the Elections Act,2017, consequently he ceases to be a member of the National Assembly of Pakistan and his seat has become vacant accordingly," the ECP order stated.

"As the respondent has made false statements and incorrect declarations, therefore he has also committed the offence of corrupt practices defined under Section 167 and 173 of the Elections Act,2017, punishable under Section 174 of the Elections Act, 2017. Office is directed to initiate legal proceedings and to take follow-up action under Section 190(2) of the Elections Act, 2017."

The verdict was announced amid tight security in the Red Zone where at least 1,100 police officials were deployed under the command of a senior superintendent of police to control the law and order situation in the area. The officer was assisted by five superintendents and six deputy superintendents of police.

The ECP had reserved its verdict in the Toshakhana case on September 19 after arguments were completed from both sides. A five-member bench of the ECP, headed by Chief Election Commissioner Sikandar Sultan Raja, heard the case.

Fawad Chaudry said the Election Commission’s attitude had been the worst. “We never pinned any hopes on it.”

“This is a shameful decision. They have disgraced the 220-million-strong-nation.”

Meanwhile, PPP Chairman Bilawal Bhutto said that the Election Commission of Pakistan has found Imran Khan guilty of corrupt practices and now stands disqualified.

“He who would spread lies about the alleged corruption of his political opponents has been caught red-handed," Bilawal said.

Moreover, PML-N Vice President Maryam Nawaz demanded Imran Khan be arrested after he was found guilty of corrupt practices.

“Pakistan's first certified liar and certified thief disqualified with irrefutable evidence of theft,” the PML-N leader said.

She alleged that the husband and wife together looted the national kitty.

In August 2022, National Assembly Speaker Raja Pervaiz Ashraf sent a reference to the ECP under Articles 62A, 63A, and 223, seeking former prime minister Imran Khan’s disqualification in the light of the Toshakhana scam, according to Geo news reports.

The disqualification reference was filed by Ali Gohar Khan, PML-N’s Mohsin Nawaz Ranjha, and five others.

The assessed value of the gifts has been put at Rs 142,042,100. The gifts were received between August 2018 and December 2021.

According to PML-N MNA Ranjha, Khan concealed information about the gift of watches by Saudi Arabia’s Crown Prince Mohammad bin Salman in his statement of assets submitted in 2018-2019.

This withholding of information is tantamount to lying which, according to Ranjha, is a crime under Section 137.

UNI GNK。 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。

三 | 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。“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。