内容来源：https://news.microsoft.com/signal/articles/4-ways-ai-is-reshaping-discovery-health-work-and-responsibility/

内容总结：

人工智能正以前所未有的深度融入日常生活，并在科学探索、医疗健康、责任伦理与工作模式四大领域引发深刻变革。微软近期推出的《再思考》系列访谈，邀请未来学家与四位内部专家，共同探讨人工智能在现实应用中的潜力、局限与责任边界。

在医疗领域，人工智能的核心价值在于辅助医生应对现代医学的复杂性。微软健康未来部门副总裁乔纳森·卡尔森指出，AI能整合海量临床资料，帮助医生进行精准诊断，推动个性化医疗发展。“治疗不应是‘AI还是医生’的二选一，”他强调，“未来必然是人与AI的协作。”

关于AI治理，微软负责任AI办公室技术顾问海沃特·特斯法耶认为，构建负责任的AI必须吸纳全球多元视角。她主导的“全球负责任AI学者计划”正汇聚来自全球南方国家的专家，以确保技术发展兼顾不同文化背景与社会需求，避免认知盲区。

在科研领域，AI正重塑科学发现的全流程。微软探索部门产品经理约翰·林克表示，AI不仅能加速实验，更能跨学科整合知识、生成假设甚至设计实验。此前，微软利用AI在十天内发现一种环保数据中心冷却分子，这传统方法需耗时数年。“这让我们看到，AI有望帮助人类更快应对气候变化、粮食安全等全球挑战。”

在工作场景中，AI正在打破职业壁垒与组织架构。微软365 Copilot及未来工作团队总经理科莱特·斯托尔鲍默指出，AI使专业知识日益“民主化”，团队正从固定的“组织架构图”转向围绕任务的“工作关系图”。她强调，持续学习与适应能力比资历更重要，“在这个时代，要么主动自我革新，要么被环境淘汰。”

尽管前路充满未知，但专家一致认为，这是一个充满可能的时代——人类正携手AI，迈向更高效、包容与创新的未来。

中文翻译：

预计阅读时间6分钟。
人工智能重塑探索、健康、工作与责任的四种方式
作者

随着人工智能融入日常生活，其影响正以具体方式日益显现：它改变着科学家的探索路径、医生的决策方式、责任与包容性议题的解决思路，以及工作组织的形态。

在《再思考》系列视频中，未来学家希内德·博维尔与四位深耕这些领域的微软研究员及主题专家展开对话。探讨聚焦于人工智能应用于现实场景时引发的变革——它创造了哪些可能性？局限何在？团队如何权衡部署这类系统时的责任与挑战？

本系列今日启动，新剧集将于1月29日、2月12日和2月26日陆续发布。以下为预告片及各集亮点。

第一集：人工智能如何通过增强人文关怀实现个性化医疗
微软研究院健康未来部门副总裁兼总经理乔纳森·卡尔森指出，人工智能在医疗领域最直接的价值在于帮助临床医生应对现代医学的极端复杂性。他认为没有医生能掌握所有专科知识、研究或指南，但人工智能能增强医生已依赖的数字工具，助其更快诊断疾病、整合相关医学知识并应用于具体患者。

"人类具有高度多样性，"他说，"我们各不相同，而医学却不得不基于平均值运作。但没有人是标准化的平均值，每个人都有独特性。因此个性化医疗的目标正是为每个个体量身定制治疗方案。"

卡尔森表示，这项技术有助于解决针对群体设计的医疗系统与个体化治疗之间的错配。例如在肿瘤学领域，许多疗法仅对部分癌症患者有效，且无法准确预测哪些患者会产生反应。

人工智能通过整合分析海量杂乱临床数据——手写笔记、PDF文件、传真、扫描影像等——帮助医生做出更精准的决策。卡尔森提到，人工智能体甚至能扮演不同专科医生的角色，综合数据以呈现更完整的病情全景，但它们不会取代人文关怀。

"人类需要人际互动、需要人性化接触、需要人类判断力，"他强调，"将问题简化为'选择人工智能还是医生'是一种错误的二元对立。未来显然是两者协同共存。"

第二集：为何负责任的人工智能需要更早吸纳多元声音
人工智能正在重塑社会，但谁来塑造人工智能？微软负责任人工智能办公室技术顾问希沃特·特斯法耶认为答案必须是：所有人。她指出，人工智能系统不仅需要技术专家的参与，更需要语言学家、社会科学家、人类学家以及全球普通用户的共同贡献。

"在人工智能系统设计和模型训练的早期阶段，我们融入的观点与视角越多元，最终成果就越完善，"她解释道。

特斯法耶结合自身在埃塞俄比亚、苏丹、肯尼亚和乌干达的成长经历谈到，幼年接触多元文化的经历塑造了她"在文化间充当翻译"的能力，这种经历持续影响着她在微软的工作方式。这种全球视野也推动了她主导的"全球负责任人工智能学者计划"，该计划汇聚来自全球南方的政策制定者、工程师和创意社群等人工智能专家，旨在完善微软的负责任人工智能项目，避免新系统与模型开发时的认知盲区。

"该计划的目标是让微软处于学习者位置，请学者们帮助我们理解：在你们所处的世界，人工智能有哪些有益应用？同时你们观察到哪些需要我们警惕的滥用现象？"她说，"我们的技术触及世界每个角落，因此必须了解这些滥用行为如何显现。"

第三集：当人工智能改变科学发现的节奏与结构
人工智能不仅加速实验室工作——智能体正通过与研究人员协同参与科学过程的每个阶段，重塑探索本身：整合海量现有研究、建立跨学科关联、提出假设甚至协助设计与实施实验。

"我们将其置于整个科学方法的框架中思考，"专注化学与材料科学的微软探索部门产品经理合伙人约翰·林克表示，"这是恰当的人工智能体与科学家在整个过程中携手共进。"

林克指出，微软始终寻找新技术能产生超常影响力的领域。当公司利用人工智能在10天内帮助发现更环保的数据中心冷却剂分子（传统方法需耗时数年）后，科学领域便成为重点方向之一。

"那是我们的'顿悟时刻'——既然我们能实现突破，不妨想象这项技术将为所有客户带来怎样的变革，"林克说。

展望未来，林克构想实验室里"每位初级科学家"都将获得"虚拟博士后团队"的支持。他认为这有望推动解决粮食安全、气候变化等全球性挑战。

"归根结底，我们需要更快解决世界难题，"他强调，"必须加速科学发现。我们相信这些工具将发挥巨大作用。"

第四集：当人工智能普及专业知识，技能比头衔更重要
微软365 Copilot及未来工作团队总经理科莱特·斯托尔鲍默指出，随着人工智能普及专业知识，传统层级结构正在扁平化，团队组建与运作模式也发生着深刻变革。

"我们团队经常讨论一种动态变化：从组织架构图转向工作关系图，"她解释道。员工不再固守静态职位，而是围绕项目动态聚合，随优先级调整而解散重组。"团队变得更流动、更敏捷，"她补充说，人工智能让人得以跨越单一学科积累专业知识。

斯托尔鲍默认为，这意味着"终身学习"已成为员工最重要的思维模式之一，并强调人工智能技能培训是企业发展的必然要求。

"这不能是事后补救措施，"她说，"在这个时代，要么主动突破自我，要么被动遭受冲击。"

她指出，这种现实正重塑企业招聘模式与职业发展路径，好奇心、适应力和实时学习能力变得比单纯的工作年限更具价值。

"当前充满不确定性，但也蕴含无限可能，"她说，"这是一个值得人们全力投入、持续学习并保持乐观的历史性时刻。"

完整系列内容将在此发布。

英文来源：

– The estimated reading time is 6 min.
4 ways AI is reshaping discovery, health, work and responsibility
Author
As AI becomes part of everyday life, its impact is increasingly showing up in concrete ways: in how scientists approach discovery, how doctors make decisions, how questions of accountability and inclusion are addressed and how work gets organized.
In the On Second Thought video series, futurist Sinead Bovell speaks with four Microsoft researchers and subject matter experts working directly in these areas. The conversations focus on what changes when AI is applied in real-world settings — what it makes possible, where its limits still are and how teams are navigating the tradeoffs that come with deploying these systems responsibly.
The series kicks off today, with new episodes coming Jan. 29, Feb. 12 and Feb. 26. Check out the trailer and highlights from each episode below.
Episode 1: How AI could make medicine more personal by enhancing human care
AI’s most immediate value in healthcare is helping clinicians navigate the sheer complexity of modern medicine, says Jonathan Carlson, vice president and managing director of Microsoft Research Health Futures. No physician can track every specialty, study or guideline, Carlson says. But AI can add to the digital tools doctors already rely on and help them more quickly diagnose an ailment, bring together relevant medical knowledge and apply it to the specific patient.
“We’re really diverse,” he says. “All of us are different. And yet medicine has to operate off of averages. Yet none of us are average. Every one of us has our idiosyncrasies, right? And so the goal of personalized medicine is to do the right thing for you, to do the right thing for me.”
The technology helps address the mismatch of treating deeply individual people with systems designed for populations, Carlson says. In oncology, for example, treatments only work for a fraction of cancer patients, with no clear way to predict who will respond.
AI can help by structuring and analyzing vast amounts of messy clinical data — written notes, PDFs, faxes, scans and more — so physicians can make more precise decisions.
AI agents can even take on different specialists’ roles to synthesize data that helps doctors see the bigger picture, he says. But they won’t replace human care.
“We as humans need human interaction, need human touch, need human judgment,” he says. “Framing the question of will it be an AI or a doctor is just a false dichotomy. It will obviously be both.”
Episode 2: Why building AI responsibly requires more voices, earlier
AI is reshaping society, but who gets to shape AI? For Hiwot Tesfaye, a technical advisor in Microsoft’s Office of Responsible AI, the answer has to be: everyone. AI systems need input not only from technologists, Tesfaye says, but from linguists, social scientists and anthropologists, as well as from everyday users around the globe.
“As many opinions and perspectives as we can incorporate as early as possible in the design of an AI system and the training of an AI model, the more voices we can incorporate, the better,” she says.
Tesfaye draws on her own experience growing up across Ethiopia, Sudan, Kenya and Uganda, where exposure to different cultures at a young age shaped her ability for “playing translator” between cultures and continues to inform how she approaches her work at Microsoft. That global lens also drives the Global Perspectives on Responsible AI Fellowship program she leads at Microsoft, which convenes AI experts from the Global South including policymakers, engineers and the creative community to better inform Microsoft’s Responsible AI program and to help avoid blind spots as new systems and models are built.
The goal of the fellowship is for Microsoft to be in the seat of the student and have the fellows “help us understand what are the beneficial uses of AI that are manifesting your part of the world, but also, what are the misuses that you’re seeing that we need to be aware of,” she says. “Our technology touches every corner of the world, so we need to be aware of how these misuses are playing out.”
Episode 3: When AI changes the pace — and structure — of scientific discovery
AI isn’t simply speeding up lab tasks — AI agents are reshaping discovery by working alongside researchers through every stage of the scientific process, synthesizing vast amounts of existing research, making connections across disciplines, generating hypotheses and even helping to design and run experiments.
“We think of it in the terms of the entire scientific method,” says John Link, a partner product manager for Microsoft Discovery, who focuses on chemistry and materials science. “It’s the right AI agents, working hand-in-hand with a scientist throughout that process.”
Microsoft looks for domains where new technologies can create outsized impact, Link says. Science stood out as one of those areas after the company used AI to help identify a new, more environmentally friendly data center coolant molecule in less than 10 days — work that would have taken years using traditional methods.
“It was our ‘aha’ moment that, hey, if we can do it, like, just imagine what this could do for all of our customers,” Link says.
Looking ahead, Link envisions labs where “every entry-level scientist” arrives already supported by “a team of virtual postdocs.” The promise, he says, is progress on global challenges like food insecurity and climate change.
“In the end, we need to solve the world’s problems faster,” he says. “We need to accelerate scientific discovery. And we think these are great tools to do it.”
Episode 4: As AI democratizes expertise, skills matter more than titles
AI is leveling the playing field at work as people in every stage and role are collectively navigating the impact of the technology on their jobs, says Colette Stallbaumer, who is the general manager of the Microsoft 365 Copilot and Future of Work teams. As AI democratizes expertise, it’s also flattening traditional hierarchies and changing how teams form and function inside organizations.
“One thing that we see is this dynamic of what we talk about in my team and describe as moving from the org chart to the work chart,” Stallbaumer says. Instead of static roles and titles, employees increasingly come together around projects, then disband and regroup as priorities shift. “Teams are fluid, more agile,” she says, adding that AI allows people to build expertise beyond a single discipline.
That means one of the most important mindsets for employees now is to be a lifelong learner, Stallbaumer says, adding that AI skilling is a business imperative.
“It can’t be an afterthought,” she says. “You’re either going to disrupt yourself or be disrupted in this environment.”
This reality is reshaping how companies hire and how careers grow, Stallbaumer says, with curiosity, adaptability and the ability to learn in real time becoming more valuable than years of experience alone.
This moment is uncertain, she says, but also full of possibility: “It’s an incredible time in history for people to lean in and learn and be optimistic.”
The full series will be available here.