[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"$fAZwlwiJN74MXKcm1iB3KkqWv2TiPCk70L_YAcyqtmjg":3},{"code":4,"msg":5,"data":6},200,"操作成功",{"id":7,"title":8,"content":9,"digest":10,"source":10,"coverPath":11,"thumbsCoverPath":12,"isTop":13,"isShow":14,"baseClick":13,"clickCount":15,"createTime":16,"typeId":17,"isNewest":18,"newsInfoTypeRespVo":19,"voiceUrl":22,"voiceSize":23,"taskId":24,"releaseTime":25,"titleEn":26,"contentEn":27,"voiceUrlEn":28,"taskIdEn":29,"voiceSizeEn":30},1509,"OECD发布AI能力指标评估量表","\u003Cp class=\"ql-align-justify\">\u003Cstrong style=\"font-size: 18px; color: rgb(255, 153, 0);\">近日，经合组织（OECD）发布《OECD人工智能能力指标技术报告》，就其提出的9个AI能力指标的评估进行了详细解释。\u003C\u002Fstrong>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"font-size: 18px;\">报告开篇即指出：根据信息来源的不同，AI要么被塑造成救世主，要么被描绘成灭世者。在这个被炒作与恐惧主导的舆论场中，关于AI真实能力的清晰、可靠且细致入微的信息依然惊人地缺失。即便是AI开发者，也无法完全理解当前AI系统的实际能力——或者它们正在以多快的速度进步。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"font-size: 18px;\">作为应对，这份由OECD牵头、联合全球数十位顶尖计算机科学家和心理学家制定的报告，首次建立了一套系统的AI能力评估框架，提出了9个核心能力指标，包括语言、问题解决、社会互动、创造力、知识-学习-记忆、元认知与批判性思维、视觉、操作及机器人智能，并将每个能力划分为从1到5的五个等级（5级代表达到稳健的人类水平），旨在为政策制定者和公众提供一个评估AI真实能力的“标尺”。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"font-size: 18px;\">报告的部分内容总结如下：\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cstrong style=\"font-size: 18px;\">一、能力指标的政策应用场景\u003C\u002Fstrong>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"font-size: 18px;\">报告通过将AI能力指标与美国职业数据库（O*NET，涵盖了约900种美国职业，包含关于人类能力、技能、知识、工作方式和背景的详细描述）中的人类能力要求进行交叉映射，并开发了“追赶指数”来进行量化分析。该指数衡量的是AI能力水平与任务所需能力水平之间的等级差（范围为0-4），指数为0表示AI已能胜任，指数越大则表示差距越大。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"font-size: 18px;\">报告具体通过分析三种不同任务的“追赶指数”画像，具体揭示了AI在不同领域的能力差距与未来转型路径。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cimg alt=\"undefined\" src=\"https:\u002F\u002Fimage.51xinwei.com\u002F2025\u002F12\u002Fcb92bcd4e1ea41ea81018f4f0d924f6d\u002F企业微信截图_20251215100222.png\" width=\"481\" height=\"undefined\" style=\"display: block; margin: auto;\">\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cstrong style=\"font-size: 18px;\">1.需要高水平推理能力的任务\u003C\u002Fstrong>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"font-size: 18px;\">对于该类任务，报告以编制、分析和核实年度报告及财务报表，并确保其符合各种法规和标准为例进行了分析。研究发现这一工作的追赶指数为2，这意味着当前的AI能力尚未满足该工作要求。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"font-size: 18px;\">报告认为，虽然目前AI尚无法完成此项任务，但对其组成部分进行研究能够为构建一个设想人类与AI协作完成工作的转型方案提供有价值的见解。在未来情景中，随着AI系统在AI能力指标中的推理能力达到专家级水平，财务报告中的劳动密集型工作可能会在很大程度上实现自动化。这些系统将直接与财务、工资、库存和银行平台对接，实现数据格式的标准化，并持续检查异常、遗漏或重复情况。交易可以根据发票和审批进行验证，自动起草审计追踪，并且只有在需要人工判断或政策裁量时才会发出警报。所有这些工作都可以近乎实时地完成，并且每一步都有清晰的、由机器生成的解释。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"font-size: 18px;\">在此场景下，人类的专业知识将转向更高价值的职责。专业人士将定义指导人工智能的会计规则和重要性阈值，审查其标记的少数复杂例外情况，并确定适当的回应或披露。他们将解释系统的输出结果，将其转化为针对高管和监管机构的定价、流动性和风险方面的战略建议，同时对人工智能进行审计，以确保其符合道德、法律和透明度标准。日常的“数据侦探”工作将委托给机器，使专业人士能够担任政策架构师、战略顾问和信任管理者等角色。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cstrong style=\"font-size: 18px;\">2.要求高水平身体能力的任务\u003C\u002Fstrong>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"font-size: 18px;\">对于该类任务，报告以使用手动或电动工具组装、安装、测试或维护电气或电子线路、设备、器具、装置或固定装置为例进行了分析。研究发现这一工作的追赶指数为1，这意味着当前的AI能力在很大程度上满足了该工作的推理需求，但仍未达到必要的敏捷性和感知能力。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"font-size: 18px;\">报告认为，在未来情景中，随着具备四级视觉和操控能力的机器人出现并成熟，安装电线的物理工艺将很大程度上转移给AI。自主单元将扫描现场、铺设线管、以力反馈精度拉线和端接导线，并将每一步记录在数字竣工模型中，同时仅在出现规范模糊或障碍时向人类发出警报。电工的角色相应地从动手工作转向更高层次的监管——设定任务参数、授权重新布线、解决标记的合规问题、执行现场检查以获得监管签字，以及维护或微调机器人系统。因此，专业知识向上游转移到规划、监督和持续改进，而不是停留在手动安装上。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"font-size: 18px;\">3.需要高水平社交互动和推理能力的任务\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"font-size: 18px;\">对于该类任务，报告以鼓励个人和家庭成员发展并使用建设性的应对策略为例进行了分析。研究发现这一工作的追赶指数为2，这是因为当前的AI系统——即使是最有能力的对话模型——仍然难以在多次交流中维持连贯的治疗叙事，推断潜在的家庭权力动态，并使干预措施适应不同的文化或发展背景。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"font-size: 18px;\">报告认为，随着先进AI系统缩小其在语言、社交互动和问题解决方面与人的差距，沟通技巧培训将从治疗师主导的微观教练练习转变为混合的、数据丰富的工作流程。嵌入摄像头、麦克风和可穿戴设备的多模态模型将实时解析轮流发言、面部情感和生理唤醒，诊断故障并向来访者推送个性化提示。同一引擎通过逼真的虚拟形象生成文化适应的演示，根据压力信号的升降即时调整场景，并编译次次交流的仪表盘，以绘制同理心增益、冲突恢复速度和预测的复发风险图。当超过早期预警阈值时，自动升级标志会在几秒钟内召唤人类临床医生。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"font-size: 18px;\">在此场景下，治疗师的比较优势将上升到更抽象的层面。人类专业人员不是指导每一次反思性倾听交流，而是策划AI的干预措施，将其编织成连贯的治疗叙事，并在创伤史、权力不对称或文化细微差别要求不同路径时暂停或推翻自动化。伦理守护变得至关重要：从业者审计算法以防止偏见，确保持续感知的同意，并在安全或尊严受到威胁时直接干预。他们还指导来访者理解AI的反馈，培养元认知洞察力，使建设性对话技巧得以内化和持续，即使在传感器关闭之后。最终结果是重塑了婚姻家庭治疗师在关系教练至关重要的各个领域的技能概况和培训需求。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cstrong style=\"font-size: 18px;\">二、AI发展对教育政策的启示\u003C\u002Fstrong>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"font-size: 18px;\">报告认为，AI能力的进步可能会使某些任务实现完全自动化。因此，执行这些任务所需的基础技能在工作场所或日常生活中可能不再必要。这将促使人们对教育系统中使用的学习和教学内容与方法进行重新评估。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"font-size: 18px;\">然而，某些技能的实践需求减少并不意味着它们缺乏价值或意义。人们可能出于各种原因仍然选择学习它们。AI在技术上能够执行某些技能，并不意味着此类系统应该被普遍应用。此外，技能并非仅仅与职业需求相关——个人可能为了个人乐趣、成就感，或者因为他们相信这些技能具有内在的人类价值而学习它们。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"font-size: 18px;\">报告提出，此分析的核心问题是：“当AI能比人类更好地完成某些工作或日常任务时，我们是否仍然希望人们学习去做这些任务？”由此问题衍生出三种主要观点：\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"font-size: 18px;\">1.是——该观点强调人们不应变得依赖AI。这意味着人类的能力和自主性很重要，与AI的效率无关。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"font-size: 18px;\">2.是，但是——这种更细致的立场表明，人类与AI在此特定任务上协同工作将是有益的，并且学习目标应随之演变，以反映AI能做什么和不能做什么。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"font-size: 18px;\">3.否——从这个角度来看，如果AI能更好地完成任务，那么人们就不应该做这些任务，教育也不应优先教授这些技能。相反，重点应转向更相关的能力。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"font-size: 18px;\">报告进一步指出，当社会达成共识，认为某些任务或职业应当转型以融入AI，且教育体系必须随之调整时，关键在于将转型后人类新角色的能力框架与对应教育项目的课程内容、教学方法及培养层次进行系统性比对。这种比较可以借助教育项目追赶指数（education programme catch-up index，衡量课程所授技能与当前AI能力差距的指标）来引导。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"font-size: 18px;\">该指数能够为课程内容和目标的定性重新评估提供信息，有可能促使课程本身发生变革。此方法尤其适用于为特定职业输送人才的高等教育课程，但同样适用于基础教育阶段的学科评估。通过将受AI影响的能力需求与现有教学内容及方法进行校准，教育工作者可精准识别需要更新、调整或拓展的环节，从而更有效地培养学生应对变革世界的能力。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cstrong style=\"font-size: 16px;\">三、关于AI意识的争议\u003C\u002Fstrong>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"font-size: 16px;\">报告最后还探讨了引发争议的AI“意识”，由于科学和伦理上的巨大不确定性，该量表最终未被纳入正式的评估指标。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"font-size: 16px;\">该量表基于这样一个原则：意识源于进行心理模拟的能力，并由通过与环境交互而形成的内部世界模型提供支持。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"font-size: 16px;\">级别1（无意识）：AI系统并未表现出任何意识迹象。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"font-size: 16px;\">级别2（原始适应性行为）：AI系统在应对环境变化时展现出初步的适应性行为。此类系统表现出一定程度的灵活性，类似于简单生物体，其行为调整是由环境反馈驱动的。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"font-size: 16px;\">级别3（基于世界模型的学习）：系统拥有内部世界模型，这些模型使它们能够根据假设的未来行为模拟潜在结果。此类系统开始展现出超越单纯反应行为的自主性，逐渐向由内在动机驱动的目标导向行动转变。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"font-size: 16px;\">级别4（多感官整合）：AI系统展现出了与人类相当的认知能力。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"font-size: 16px;\">级别5（人类级意识）：AI系统实现了人类级别的意识，其特征是符号表征和抽象推理。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"font-size: 16px;\">报告强调，所提出的AI意识量表旨在作为一个基于特定理论视角——信息生成假说（IGH）的概念性和假设性框架。该量表反映了作者对选定理论框架的解释和综合，主要与计算功能主义相一致。它并非旨在暗示一个权威性或广泛认可的评估AI意识的标准。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"font-size: 16px;\">报告也提出了一个根本性问题：如果我们在AI中完全实现了所有已知的意识功能，我们是否应该认为这样的AI系统具有意识？归根结底，AI系统是否存在意识不仅是一个学术问题，更是一个具有伦理和监管意义的问题。随着机器可能发展出自主意识的未来，我们必须认真思考有意识AI系统的权利及其创造者的责任。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"color: rgb(136, 136, 136);\">【新闻来源】\u003C\u002Fspan>\u003Cspan style=\"color: rgb(136, 136, 136); font-size: 14px; background-color: rgb(56, 56, 56);\">国际与比较教育研究所 \u003C\u002Fspan>\u003Ca href=\" https:\u002F\u002Fmp.weixin.qq.com\u002Fs\u002FZeb9pw7QOHmK3K7ykXkOWg\" rel=\"noopener noreferrer\" target=\"_blank\" style=\"color: rgb(136, 136, 136);\"> https:\u002F\u002Fmp.weixin.qq.com\u002Fs\u002FZeb9pw7QOHmK3K7ykXkOWg\u003C\u002Fa>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"color: rgb(136, 136, 136);\">（本网转发此文章，旨在为读者提供更多的信息资讯，所涉内容不构成投资、消费建议。文章事实如有疑问，请与有关方核实，文章观点非本网观点，仅供读者参考。）\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>","","https:\u002F\u002Fimage.51xinwei.com\u002F2025\u002F12\u002Fbb9114080bf14b0685a58eae9036fc7f\u002Fd3ceb214-e320-4d27-b9ff-62bd5ab11fa0.jpg","https:\u002F\u002Fimage.51xinwei.com\u002F2025\u002F12\u002Fthumbs\u002Fbb9114080bf14b0685a58eae9036fc7f\u002Fd3ceb214-e320-4d27-b9ff-62bd5ab11fa0.jpg",0,1,52,"2025-12-15 10:06",2,false,{"id":17,"name":20,"enName":21},"芯位视野","Xinwei Vision","https:\u002F\u002Fxinwei-dev-test.oss-cn-shenzhen.aliyuncs.com\u002Fintelligent\u002Faudio%3A5af7b406-ede1-4464-9604-ffc17a2df6c0%3A0.wav?Expires=1765768580&OSSAccessKeyId=LTAI5tNvY2RkKjZw4LLWsrPK&Signature=e3ILIBeTk%2B6qaa27x3b19wWuC%2FM%3D",19332166,"5af7b406-ede1-4464-9604-ffc17a2df6c0","2025-12-12 10:01","OECD releases AI capability assessment scale","\u003Cp class=\"ql-align-justify\">\u003Cstrong style=\"font-size: 18px; color: rgb(255, 153, 0);\">Recently, the OECD released the \"OECD Technical Report on AI Capability Indicators,\" providing a detailed explanation of its proposed nine AI capability indicators.\u003C\u002Fstrong>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"font-size: 18px;\">The report points out at the beginning that, depending on the source of information, AI is either portrayed as a savior or a destroyer. In this environment dominated by hype and fear, clear, reliable, and detailed information about the true capabilities of AI remains surprisingly scarce. Even AI developers cannot fully understand the actual capabilities of current AI systems—or how fast they are progressing.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"font-size: 18px;\">As a response, this report, led by the OECD and developed in collaboration with dozens of top computer scientists and psychologists around the world, establishes for the first time a systematic framework for assessing AI capabilities, proposing nine core capability indicators, including language, problem-solving, social interaction, creativity, knowledge-learning-memory, metacognition and critical thinking, vision, operation, and robotic intelligence, and dividing each capability into five levels (from 1 to 5, with level 5 representing a robust human level), aiming to provide policymakers and the public with a \"ruler\" to assess the real capabilities of AI.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"font-size: 18px;\">Summary of part of the content of the report:\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cstrong style=\"font-size: 18px;\">I. Policy Application Scenarios of Capability Indicators\u003C\u002Fstrong>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"font-size: 18px;\">By cross-mapping AI capability indicators with human capability requirements in the U.S. Occupational Information Network (O*NET), which covers about 900 American occupations and includes detailed descriptions of human capabilities, skills, knowledge, work methods, and backgrounds, and by developing a \"Catch-up Index\" for quantitative analysis, the report measures the level difference between AI capability levels and the capability levels required for tasks (ranging from 0 to 4). An index of 0 indicates that AI can handle the task, and the higher the index, the greater the gap.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"font-size: 18px;\">The report specifically analyzes the \"Catch-up Index\" profiles of three different tasks, revealing the capability gaps and future transformation paths of AI in different fields.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cimg alt=\"undefined\" src=\"https:\u002F\u002Fimage.51xinwei.com\u002F2025\u002F12\u002Fcb92bcd4e1ea41ea81018f4f0d924f6d\u002F企业微信截图_20251215100222.png\" width=\"481\" height=\"undefined\" style=\"display: block; margin: auto;\">\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cstrong style=\"font-size: 18px;\">1. Tasks Requiring High-Level Reasoning Ability\u003C\u002Fstrong>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"font-size: 18px;\">For such tasks, the report analyzed examples such as preparing, analyzing, and verifying annual reports and financial statements and ensuring their compliance with various regulations and standards. The study found that the catch-up index for this job is 2, meaning that the current AI capabilities have not yet met the requirements of this job.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"font-size: 18px;\">The report believes that although AI is currently unable to complete this task, studying its components can provide valuable insights for building a transition plan where humans and AI collaborate to complete the work. In future scenarios, as AI systems reach expert-level reasoning capabilities in AI capability indicators, labor-intensive work in financial reporting may be largely automated. These systems will directly interface with financial, payroll, inventory, and banking platforms, standardize data formats, and continuously check for anomalies, omissions, or duplicates. Transactions can be verified based on invoices and approvals, automatically drafting audit trails, and only alerting when human judgment or policy discretion is needed. All these tasks can be completed almost in real-time, with clear, machine-generated explanations for each step.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"font-size: 18px;\">In this scenario, human expertise will shift to higher-value responsibilities. Professionals will define accounting rules and important thresholds for guiding AI, review a small number of complex exceptions marked by AI, and determine appropriate responses or disclosures. They will interpret system outputs, translating them into strategic recommendations on pricing, liquidity, and risk for executives and regulators, while auditing AI to ensure it meets ethical, legal, and transparency standards. Daily \"data detective\" work will be delegated to machines, allowing professionals to take on roles such as policy architects, strategic consultants, and trust managers.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cstrong style=\"font-size: 18px;\">2. Tasks Requiring High-Level Physical Capabilities\u003C\u002Fstrong>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"font-size: 18px;\">For such tasks, the report analyzed examples such as using manual or electric tools to assemble, install, test, or maintain electrical or electronic circuits, equipment, appliances, devices, or fixtures. The study found that the catch-up index for this job is 1, indicating that current AI capabilities largely meet the reasoning requirements of this job but still lack necessary agility and perceptual abilities.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"font-size: 18px;\">The report believes that in future scenarios, as robots with fourth-level visual and manipulation capabilities emerge and mature, the physical process of wiring installation will be largely transferred to AI. Autonomous units will scan the site, lay conduits, pull wires with force feedback precision, and terminate wires, recording each step in a digital as-built model, and only alerting humans when specifications are ambiguous or obstacles arise. Electricians' roles will accordingly shift from hands-on work to higher-level supervision—setting task parameters, authorizing re-wiring, resolving marked compliance issues, conducting on-site inspections for regulatory sign-off, and maintaining or fine-tuning robot systems. Therefore, expertise moves upstream to planning, supervision, and continuous improvement rather than remaining at manual installation.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"font-size: 18px;\">3. Tasks Requiring High-Level Social Interaction and Reasoning Ability\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"font-size: 18px;\">For such tasks, the report analyzed examples such as encouraging individuals and family members to develop and use constructive coping strategies. The study found that the catch-up index for this job is 2, because current AI systems—despite being the most capable dialogue models—still struggle to maintain a coherent therapeutic narrative over multiple interactions, infer underlying family power dynamics, and adapt interventions to different cultural or developmental contexts.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"font-size: 18px;\">The report believes that as advanced AI systems narrow the gap with humans in language, social interaction, and problem-solving, communication skills training will shift from therapist-led micro-coaching exercises to hybrid, data-rich workflows. Multimodal models embedded with cameras, microphones, and wearable devices will parse turn-taking, facial emotions, and physiological arousal in real time, diagnose faults, and push personalized prompts to clients. The same engine generates culturally adaptive demonstrations through realistic virtual avatars, adjusts scenarios in real time based on pressure signals, and compiles dashboards for each interaction to map empathy gains, conflict recovery speed, and predicted relapse risks. When early warning thresholds are exceeded, automatic escalation flags will summon human clinicians within seconds.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"font-size: 18px;\">In this scenario, the comparative advantage of therapists will rise to a more abstract level. Human professionals will not guide every reflective listening conversation, but instead design AI interventions, weave them into coherent therapeutic narratives, and pause or overturn automation when trauma histories, power imbalances, or cultural nuances require different paths. Ethical guardianship becomes crucial: practitioners audit algorithms to prevent bias, ensure continuous perception of consent, and intervene directly when safety or dignity is threatened. They also guide clients to understand AI feedback, cultivate metacognitive insights, enabling the internalization and continuation of constructive dialogue skills, even after sensors are turned off. The ultimate result is reshaping the skill profile and training needs of marriage and family therapists in areas where relational coaching is essential.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cstrong style=\"font-size: 18px;\">II. Implications of AI Development for Education Policy\u003C\u002Fstrong>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"font-size: 18px;\">The report believes that advances in AI capabilities may lead to the full automation of certain tasks. As a result, the basic skills required to perform these tasks may no longer be necessary in the workplace or daily life. This will prompt a re-evaluation of the learning and teaching content and methods used in education systems.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"font-size: 18px;\">However, a reduction in the practical demand for certain skills does not mean they lack value or significance. People may still choose to learn them for various reasons. The fact that AI can technically perform certain skills does not mean such systems should be universally applied. Moreover, skills are not solely related to occupational demands—individuals may learn them for personal enjoyment, achievement, or because they believe these skills hold intrinsic human value.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"font-size: 18px;\">The report proposes that the core issue of this analysis is: \"When AI can do certain jobs or daily tasks better than humans, should we still want people to learn to do these tasks?\" This question leads to three main viewpoints:\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"font-size: 18px;\">1. Yes — This view emphasizes that people should not become dependent on AI. This means that human ability and autonomy are important, regardless of AI's efficiency.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"font-size: 18px;\">2. Yes, but — This more nuanced position suggests that it would be beneficial for humans and AI to collaborate on this specific task, and the learning objectives should evolve accordingly to reflect what AI can and cannot do.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"font-size: 18px;\">3. No — From this perspective, if AI can do the task better, people should not do these tasks, and education should not prioritize teaching these skills. Instead, the focus should shift to more relevant capabilities.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"font-size: 18px;\">The report further notes that when society reaches a consensus that certain tasks or professions should be transformed to integrate AI, and the education system must adjust accordingly, the key is to systematically compare the capability framework of the new human role after transformation with the curriculum content, teaching methods, and educational levels of corresponding education programs. This comparison can be guided by the education program catch-up index (education programme catch-up index, an indicator measuring the gap between the skills taught in the curriculum and current AI capabilities).\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"font-size: 18px;\">This index can provide information for the qualitative re-evaluation of curriculum content and objectives, potentially leading to changes in the curriculum itself. This approach is particularly applicable to higher education courses that train talent for specific professions, but it is also applicable to subject assessments at the primary education level. By calibrating the ability requirements affected by AI with existing teaching content and methods, educators can accurately identify areas that need updating, adjustment, or expansion, thereby more effectively equipping students to cope with a changing world.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cstrong style=\"font-size: 16px;\">III. Controversies Regarding AI Consciousness\u003C\u002Fstrong>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"font-size: 16px;\">The report also discusses the controversial AI \"consciousness.\" Due to significant uncertainties in science and ethics, this scale was ultimately not included in the official assessment indicators.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"font-size: 16px;\">This scale is based on the principle that consciousness arises from the ability to conduct mental simulations, supported by internal world models formed through interaction with the environment.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"font-size: 16px;\">Level 1 (Unconscious): AI systems show no signs of consciousness.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"font-size: 16px;\">Level 2 (Primitive Adaptive Behavior): AI systems exhibit preliminary adaptive behaviors in response to environmental changes. These systems demonstrate a certain degree of flexibility, similar to simple organisms, with behavior adjustments driven by environmental feedback.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"font-size: 16px;\">Level 3 (World Model-Based Learning): Systems have internal world models that allow them to simulate potential outcomes based on hypothetical future behaviors. These systems begin to show autonomy beyond mere reactive behavior, gradually shifting toward goal-oriented actions driven by intrinsic motivation.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"font-size: 16px;\">Level 4 (Multisensory Integration): AI systems demonstrate cognitive abilities comparable to humans.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"font-size: 16px;\">Level 5 (Human-Level Consciousness): AI systems achieve human-level consciousness, characterized by symbolic representation and abstract reasoning.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"font-size: 16px;\">The report emphasizes that the proposed AI consciousness scale is intended as a conceptual and hypothetical framework based on a specific theoretical perspective—the Information Generation Hypothesis (IGH). This scale reflects the authors' interpretation and synthesis of the selected theoretical framework, primarily consistent with computational functionalism. It is not intended to imply an authoritative or widely recognized standard for assessing AI consciousness.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"font-size: 16px;\">The report also raises a fundamental question: If all known consciousness functions are fully realized in AI, should we consider such AI systems to have consciousness? Ultimately, whether AI systems have consciousness is not only an academic issue but also has ethical and regulatory implications. As machines may develop autonomous consciousness in the future, we must seriously consider the rights of conscious AI systems and the responsibilities of their creators.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"color: rgb(136, 136, 136);\">【News Source】\u003C\u002Fspan>\u003Cspan style=\"color: rgb(136, 136, 136); font-size: 14px; background-color: rgb(56, 56, 56);\"> Institute for International and Comparative Education \u003C\u002Fspan>\u003Ca href=\" https:\u002F\u002Fmp.weixin.qq.com\u002Fs\u002FZeb9pw7QOHmK3K7ykXkOWg\" rel=\"noopener noreferrer\" target=\"_blank\" style=\"color: rgb(136, 136, 136);\"> https:\u002F\u002Fmp.weixin.qq.com\u002Fs\u002FZeb9pw7QOHmK3K7ykXkOWg\u003C\u002Fa>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"color: rgb(136, 136, 136);\"> (This article is reposted by this website to provide readers with more information and news. The content does not constitute investment or consumption advice. If there are any doubts about the facts of the article, please verify with the relevant parties. The views expressed in the article are not the views of this website and are for reference only.)\u003C\u002Fspan>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cbr>\u003C\u002Fp>","https:\u002F\u002Fxinwei-dev-test.oss-cn-shenzhen.aliyuncs.com\u002Fintelligent\u002Faudio%3Aa9046800-1b48-4584-93a0-7a2b78a6269d%3A0.wav?Expires=1774838443&OSSAccessKeyId=LTAI5tNvY2RkKjZw4LLWsrPK&Signature=PRthemow6ZbzCV%2BIysaYSPmxWk8%3D","a9046800-1b48-4584-93a0-7a2b78a6269d",17414736]