一颗有心的人工智能-人工智能造福人类的好处

For several decades artificial intelligence (AI) has looked set to change the world as we know it. But what societal benefits will this technological revolution bring and will they be evenly spread?

几十年来，人工智能(AI)似乎已准备改变我们所知道的世界。 但是，这场技术革命将带来什么社会利益，并且它们会平均分配吗？

Artificial intelligence will have a larger impact on the world than the internet revolution has had so far, according to 62% of global CEOs in PwC’s 22nd Global CEO Survey, with 79% saying that AI is good for society. Google CEO Sundar Pichai has stated that AI is “more profound than fire or electricity.”

一个 rtificial情报将对世界比互联网革命，迄今已经有较大的影响，根据全球CEO中62％ 普华永道22日全球CEO调查 ，有79％的人说AI是对社会有利。 谷歌首席执行官桑达尔·皮查伊 ( Sundar Pichai )表示，人工智能“比火或电更深刻”。

Since the 1950s the spectre of a predominant AI has been popularised in science fiction books and films. The idea of all-powerful, intelligent machines is not just confined to books and art. The University of Cambridge’s Centre for the Study of Existential Risk, co-founded by Astrophysicist Martin Rees with high-profile advisers including Elon Musk, studies the very real dangers that AI create for humanity.

自1950年代以来，主要的AI幽灵在科幻小说和电影中得到了普及。 全能智能机器的想法不仅限于书籍和艺术品。 由天体物理学家Martin Rees与包括Elon Musk在内的知名顾问共同创立的剑桥大学存在风险研究中心研究了AI对人类造成的真正危险。

“[AI] more profound than fire or electricity”

“ [AI]比火或电更深刻”

This much-hyped technology has had a fair share of disillusionment including two so-called ‘AI winters’, periods when funding and interest in AI plummeted dramatically. However, recent advances in big data, data science and cloud computing make it seem that this time is different. Tata Sons Chairman Natarajan Chandrasekaran said, “Suddenly, cloud computing has made possible the real-time collection of infinite amounts of data. This opens up the possibilities for AI." The recent rise of edge computing, 5G and quantum computing, which rely on or support the growth of machine learning, a subset of AI, back up this claim.

这项被大肆宣传的技术在幻灭中占了相当大的份额，其中包括两个所谓的“人工智能冬天”，在此期间，人工智能的资金和兴趣急剧下降。 但是，大数据，数据科学和云计算方面的最新进展似乎使这次不同了。 Tata Sons董事长Natarajan Chandrasekaran表示：“突然之间，云计算使实时收集无限数量的数据成为可能。 边缘计算，5G和量子计算的最新兴起支持了这一主张，而边缘计算，5G和量子计算的出现依赖或支持作为AI的一部分的机器学习的发展。

“We’re past Pong, we’re maybe at PacMan by now”

“我们过去了Pong，现在我们也许在吃豆人”

AI is currently being put to work in three main areas: automating tasks that humans would rather not do or do badly, personalising content suggestions for online consumers and recognising patterns to classify objects and predict future outcomes. This has driven a steep increase in investor interest with annual private investment in AI quadrupling from 2015 to 2019, reaching over $35bn. As Graphcore CEO Nigel Toon stated, “We’re past Pong, we’re maybe at PacMan by now.” Investment in the sector has not been limited to the rich world. Google opened its first Africa Artificial Intelligence Centre in Ghana in 2019 and organisations such as Knowledge 4 All are working to build up AI engineering talent globally.

人工智能目前在三个主要领域发挥作用：自动化人类不愿做或做得不好的任务，为在线消费者提供个性化的内容建议以及识别模式以对对象进行分类并预测未来的结果。 从2015年到2019年，人工智能领域的年度私人投资翻了两番，达到350亿美元以上，这带动了投资者兴趣的急剧增加。 正如Graphcore首席执行官Nigel Toon所说：“我们过去了Pong，现在我们可能在PacMan。” 在该领域的投资不仅限于富裕国家。 谷歌于2019年在加纳开设了第一家非洲人工智能中心，像Knowledge 4 All这样的组织正在努力在全球范围内培养AI工程人才。

永远的人工智能 (AI for Good)

While predicting whether someone will default on their car insurance is certainly useful for underwriters, being able to identify a malignant from a benign tumour is life-changing. So what are the most impactful ways to use AI and machine learning? Those who champion AI suggest it could solve international crises, bring an end to poverty and help prevent catastrophe. There are a growing number of organisations seeking to use the power of AI for good, creating positive social and environmental benefits. Here are seven examples:

虽然预测某人是否会拖欠其汽车保险无疑对保险人有用，但能够从良性肿瘤中识别出恶性肿瘤改变了人们的生活。 那么使用AI和机器学习最有影响力的方式是什么？ 那些拥护人工智能的人表示，它可以解决国际危机，消除贫困并帮助预防灾难。 越来越多的组织寻求善用AI的力量，从而创造积极的社会和环境效益。 这是七个示例：

1. Spotting signs of abuse

   发现滥用迹象

The rAInbow chatbot from AI for Good helps to spot the signs of abuse, judge what’s healthy and unhealthy behaviour, and provide resources that can help people facing gender-based violence.

AI for Good的rAInbow聊天机器人可帮助发现虐待迹象，判断健康和不健康行为，并提供可帮助面临基于性别暴力的人们的资源。

2. Fighting the spread of deadly diseases

2.与致命疾病的传播作斗争

Using machine learning, Zzapp Malaria’s software app optimises intervention strategies that target mosquitoes, prioritising the use of scarce resources and maximising impact. The map-based mobile app guides field workers to streamline execution and monitor progress.

Zzapp Malaria的软件应用程序使用机器学习来优化针对蚊子的干预策略，优先利用稀缺资源并最大程度地发挥影响。 基于地图的移动应用程序可指导现场工作人员简化执行过程并监控进度。

3. Measuring the social impact of development initiatives

3.衡量发展举措的社会影响

The UN uses machine learning in its Radio Content Analysis Tool to accelerate sustainable development solutions in Uganda by using speech recognition to leverage public radio content as a source of information on issues relevant to sustainable development.

联合国在其无线电内容分析工具中使用机器学习，通过语音识别来利用公共广播内容作为有关可持续发展问题的信息源，从而加快乌干达的可持续发展解决方案。

4. Boosting education

4.促进教育

UNICEF Innovation is applying Deep Learning techniques to map every school in the world. The tool uses high-resolution satellite imagery which is visualized through an online platform to help identify where gaps and information needs are. This helps national governments optimise their education systems, assess vulnerabilities and enhance emergency crisis responses.

联合国儿童基金会的创新活动正在应用深度学习技术来绘制世界上每所学校的地图。 该工具使用高分辨率卫星图像，该图像通过在线平台可视化，以帮助确定差距和信息需求在哪里。 这有助于各国政府优化其教育系统，评估漏洞并增强紧急危机应对能力。

5. Enhancing healthcare assessments and delivery

5.加强医疗保健评估和交付

Össur, which focusses on prosthetic, osteoarthritis and injury support solutions, has developed artificial limbs to provide greater comfort by using machine learning to adjust the mobility equipment according to the user’s unique gait.

Össur专注于假肢，骨关节炎和损伤支持解决方案，已开发出人造肢体，通过使用机器学习根据用户的独特步态调整移动设备来提供更大的舒适度。

6. Improving agriculture production

6.改善农业生产

mCrops embeds machine learning in its diagnostic tools which identify viral crop diseases in cassava crops by taking pest and symptom measurements using a mobile device.

mCrops在其诊断工具中嵌入了机器学习功能，该工具通过使用移动设备进行有害生物和症状测量来识别木薯作物中的病毒性作物疾病。

7. Mitigating further climate change

7.减轻进一步的气候变化

Rainforest Connection detects illegal logging over 2,500 sq km of the rainforest by using acoustic monitoring and AI. By feeding audio signals from the rainforest into Google’s open-source machine learning framework, TensorFlow, Rainforest Connection can locate sounds of illegal activity.

Rainforest Connection使用声音监控和AI检测到2,500平方公里的热带雨林非法采伐。 通过将来自热带雨林的音频信号输入到Google的开源机器学习框架TensorFlow中，热带雨林连接可以定位非法活动的声音。

促进增长 (Catalysing growth)

These examples are just some of many which have shown that AI can be used as a power for good. As more entrepreneurs seek to solve challenges around the world using AI-enabled approaches, there should also be a concerted effort at an ecosystem level to support these founders and businesses. Below are five calls to action to achieve this growth and maximise the impact of AI for Good projects.

这些例子只是许多例子，这些例子表明AI可以永远用作力量。 随着越来越多的企业家寻求使用支持AI的方法来解决全球范围内的挑战，还应该在生态系统层面上共同努力，为这些创始人和企业提供支持。 以下是实现这一增长并最大程度地发挥AI for Good项目的影响的五个行动号召。

与更广泛的Tech for Good运动联系 (Connect with the wider Tech for Good movement)

With the growth of impact investing, Tech for Good has been able to go by the name of ‘high-growth impact’. However, more broadly, Tech for Good companies lack a rigid definition and set of standards, causing the communication of the ‘sector’ to suffer from a lack of clarity. AI for Good can benefit from the enthusiasm Tech for Good has created while retaining unique metrics for evaluating the social impact and viability of its initiatives. This horizontal network growth can be particularly effective for early-stage businesses for which lessons learned and best practice sharing can be largely sector agnostic.

随着影响力投资的增长，Tech for Good能够被称为“高增长影响力”。 但是，从更广泛的意义上讲，“以技术换取善”公司缺乏严格的定义和标准集，导致“部门”的沟通缺乏明确性。 AI for Good可以从Tech for Good创建的热情中受益，同时保留用于评估其计划的社会影响和可行性的独特指标。 这种水平的网络增长对于早期业务尤其有效，因为这些业务的经验教训和最佳实践共享在很大程度上与部门无关。

开发展示成功案例的平台 (Develop platforms which showcase success stories)

Competitions and growth programmes like Tech Nation’s Applied AI are fantastic ways to shine a light on successful companies. These platforms typically also support ventures to access investment, grow their customer base and boost hiring through job boards. By showcasing successful businesses and the viability of mission-driven AI companies, the next wave of entrepreneurs will be inspired to create solutions to challenges they face in communities around the world.

诸如Tech Nation的Applied AI之类的竞赛和成长计划是向成功的公司发光的绝佳方法。 这些平台通常还支持企业获得投资，扩大他们的客户群并通过工作委员会促进雇用。 通过展示成功的业务和任务驱动的AI公司的生存能力，下一波企业家将受到启发，为他们在世界各地的社区中面临的挑战提供解决方案。

建立紧密的创始人和天使投资人网络 (Build tight networks of founders and angel investors)

Businesses that seek venture capital and go through the funding cycle gain access to the networks that investors offer. For Tech for Good firms, this often comes with the additional challenge of explaining their purpose-and-profit business model. Traditional investors who obtain board seats and voting rights may try to direct the business towards maximising profits at the expense of the mission-driven impact sought by its founders. These problems are felt most acutely at the early stage of the business life cycle and can be mitigated by creating an ecosystem of like-minded angel investors in the AI for Good sphere.

寻求风险资本并经历融资周期的企业可以访问投资者提供的网络。 对于Tech for Good公司而言，这通常伴随着解释其宗旨和利润业务模型的额外挑战。 获得董事会席位和投票权的传统投资者可能会试图以牺牲创始人的使命驱动型影响为代价，将业务引导至利润最大化。 这些问题在业务生命周期的早期阶段最为明显，可以通过在AI for Good领域中创建由志趣相投的天使投资者组成的生态系统来缓解。

使用现有行业内的现有网络 (Use existing networks within established sectors)

AI is a technology, not a sector. Therefore it is crucial for any companies creating AI-enabled solutions to establish sector-specific networks in traditional sectors such as healthcare and education. This vertical network building is most useful for growth-stage ventures that seek to compete with incumbents, expand internationally and create meaningful partnerships in their industry. Unlike established sectors, it is often difficult for Tech for Good businesses to explain their place in the ecosystem and this would support their narrative and help them to appear more credible.

人工智能是一种技术，而不是部门。 因此，对于任何创建支持AI的解决方案的公司来说，在医疗保健和教育等传统行业中建立针对特定行业的网络都是至关重要的。 这种垂直网络的构建对于寻求与现有企业竞争，在国际上扩展并在其行业中建立有意义的合作伙伴关系的成长型企业最有用。 与既有部门不同，“以科技换好产品”的企业通常很难解释其在生态系统中的位置，这将支持它们的叙述并帮助它们显得更可信。

解锁试点项目的数据 (Unlock data for pilot projects)

AI and machine learning require large amounts of good quality and relevant data to build accurate models of the real world. On top of this, data used in AI for Good initiatives tends to be expensive and hard to access because of its value to current owners and sensitivity. Large owners of data, including governments, healthcare providers, schools and telecommunications companies, should unlock small parts of the data they hold for use in pilot projects to allow companies to demonstrate the value AI-enabled projects can bring, both in terms of value-added services and social impact. Doing so would further open up the door for large-scale collaboration.

人工智能和机器学习需要大量的高质量和相关数据来构建真实世界的准确模型。 最重要的是，“人工智能造福人”计划中使用的数据由于对当前所有者的价值和敏感性而趋向于昂贵且难以访问。 大型数据所有者，包括政府，医疗保健提供者，学校和电信公司，应解锁其持有的一小部分数据以供试点项目使用，以使公司能够展示基于AI的项目可以带来的价值，无论是在价值方面，增加服务和社会影响。 这样做将进一步打开大规模合作的大门。

Undoubtedly, AI is a powerful technology that can be used to efficiently and cheaply solve some of society’s challenges. Governance of AI systems will play a significant part in ensuring that the technology is used appropriately. Ultimately, the flow of capital and the role of regulation will decide how the field of AI develops. However, researchers and practitioners can play their part in making it an accessible and open-minded pursuit in both academia and business. Noticeable gains have been made where humans and AI algorithms collaborate. Partners Martin Casado and Matt Bornstein at Andreessen Horowitz, a venture capital firm, reckon that “the need for human intervention will likely decline as the performance of AI models improves. It’s unlikely, though, that humans will be cut out of the loop entirely.”

毫无疑问，人工智能是一项强大的技术，可用于高效，廉价地解决社会的一些挑战。 人工智能系统的治理将在确保正确使用该技术方面发挥重要作用。 最终，资本的流动和监管的作用将决定AI领域的发展。 但是，研究人员和从业人员可以发挥自己的作用，使之成为学术界和企业界均可访问和开放的追求。 在人类和AI算法协作的地方，已经取得了显著成就。 风险投资公司Andreessen Horowitz的合伙人Martin Casado和Matt Bornstein认为，“随着AI模型性能的提高，人工干预的需求可能会下降。 不过，不太可能将人类完全淘汰。”

Entrepreneurs will continue to seek underserved areas and create adapted AI solutions. Government authorities should welcome such approaches and seek to understand the benefits that AI can bring. In doing so they might be more willing to open up the data that they hold. Without conscious global efforts across the ecosystem, the real winner will be AI itself as it moves unseen in the back-end eating more data, growing ever smarter and taking a bigger seat at the decision-making table.

企业家将继续寻找服务不足的地区，并创建适应性强的AI解决方案。 政府当局应该欢迎这种方法，并试图理解人工智能可以带来的好处。 这样一来，他们可能会更愿意开放他们持有的数据。 没有整个生态系统的有意识的全球努力，真正的赢家将是AI本身，因为它在后端看不见，它吞噬了更多数据，变得越来越聪明，并且在决策桌上占据了更大的席位。