Google's secret use of 1.6 million patients of private data should not-Google, privacy, artificial intelligence, information on AlphaGo-IT

When Google its DeepMind London companies will be able to get Royal Tzu operating three hospitals NHS Trust Fund (Barnet, and Chase Farm and the Royal Free) 1.6 million patients NHS (United Kingdom national health service) when the news of the record was released, alarm bells rang.

This is not only because the British are fiercely guarded their private medical records; not just because their tentacles into every corner of our lives of giant squid and Google wants to "organize the world's information"; not only because patients are less likely to agree that Google acquired this information.

DeepMind ▲ Google AlphaGo and world champion Li Shishi chess game broadcast, AlphaGo win

For many people, the problem is that these fears and artificial intelligence (AI) the concept of weaving together. DeepMind is not a normal company, it is a company that focuses on artificial intelligence, technology developed to demonstrate intelligent reasoning abilities.

Last year, the company's engineers have published a paper saying the company developed a copy of "professional human video game tester" program. In March, Google DeepMind developed a master with a 3,000-year history of the Chinese board game go, which made history until then, go the huge number of possible steps, it is believed that currently are not able to master the game of technology. In the game as a milestone in the calculation of, the company's AlphaGo 4-1 defeated the World Chess Champion.

Before the illness cure

Now, such a company has all these people's detailed database of medical records, privacy, although anonymous, including HIV infection, drug overdoses and abortion information of the past. DeepMind says it needs the data to try to prevent acute kidney injury in hospital medical alarm signal.

Some fear is that DeepMind database may be permitted for purposes outside of the provisions. Public facts for NHS patients ' privacy is not guaranteed has been no stranger to--2014 years, because of medical confidentiality concerns may be at risk, the Government was forced to suspend and then scaled back their proposal to create a single English-language medical database.

DeepMind did not hide its work with the NHS, the company announced in February that it is working with the medical service cooperation in developing an application called Streams to help doctors and nurses monitor kidney disease patients. It did not release data, including patient's past medical history, is the degree of access. DeepMind got not just thousands of patients with renal injury data, but all of these hospital patient records. That's millions of classified documents.

The company said it needs the entire patient database, to make the Streams available. Supporters of such databases claiming to have this kind of data, you can quickly develop powerful software package for diagnosing diseases.

New Scientist magazine received its DeepMind and data sharing agreements between the NHS, which discloses what information will be provided to DeepMind. The Google company's skill is to recognize complex patterns in large amounts of data--and the NHS is such a "deep learning" a gold mine.

This treasure trove is filled with hospital daily activity logs, such as the location and status of the patient records--when and who visited them. DeepMind also records the patient's pathology and radiology.

In addition to real-time data, DeepMind also has the right to use intensive care and accident and Emergency Department of history. In theory, DeepMind tinker with these information can be based on these more than any individual digestion and analysis of data developed ability to forecast.

By comparing patient data DeepMind may be able to predict a person in some there are no obvious symptoms in the early stages of the disease. This is the Holy Grail of medical fields: not when people are sick patients, but to heal them before they were ill.

Utopia? It might be. The vision behind these technologies there is the crux of the dilemma of the age in which we live. Google, Facebook and other company's survival depends on this fact: our own right to privacy in Exchange for use of the data and development of new technologies.

An "intelligence explosion"

And Apple, Google's reputation is building in the area of medical applications. In fact, academia will be machine learning in health care is not news. However, access to these data being handed over to one of the biggest and most powerful companies in the world. This brings us to a question, that is, the company may soon become the largest company in the NHS health analysis, resulting in a de facto monopoly.

AI represents something new, a program that hopes to improve itself and will soon exceed human intelligence. This is known as an "intelligence explosion"--a human ushered the nodes of its own destruction. We have at this point is still some distance away. Even if there is such a target price, and still nobody developed a respect for social and moral norms of the machine. In humans, this is very difficult.

Some might say such deductions are absurd. After all, Microsoft's Twitter chat robot Tay into a racist and shout a few hours before the genocide, she was then closed. But, as originally invested in DeepMind inventor Yilong·masike said, was the "Terminator" technology concerns drove him to warn of its dangers.

May sound business reasons, DeepMind people operate under the watchful eye of. But it usually brings more questions than given answers. Google in 2014 to ￡ 400 million takeover DeepMind technology of creation of artificial intelligence, the Ethics Committee is still one of the biggest mysteries, both companies have declined to say who the Committee.

Artificial intelligence data needs to be learning, so Google DeepMind to absorb all of these records. So why the secrecy? If the patient has been informed how things worked and why, they can make an informed choice. If they think Google is in the NHS the potential risk of a monopoly on the key technologies of the new miss its benefits, let us debate it. But if the company does not make any explanation, is conducted in secrecy, the public would certainly not endorse such a plan.

谷歌秘密使用160万病人隐私数据，不应该 - 谷歌,隐私,人工智能,AlphaGo - IT资讯

当谷歌旗下位于伦敦的公司DeepMind将可以获取皇家慈济NHS信托基金运营的三家医院（Barnet、Chase Farm和the Royal Free）的160万病人的NHS（英国国家医疗服务体系）记录的消息被公布出来时，警钟敲响了。

这不只是因为英国人激烈地守卫他们私密的医疗记录；也不只是因为把触角伸进我们生活的各个角落的巨型章鱼公司谷歌想要“组织全世界的信息”；更不只是因为病人不太可能同意谷歌获得这些信息。

▲谷歌DeepMind的AlphaGo与世界冠军李世石的围棋比赛的直播，AlphaGo最终获胜

对许多人而言，问题在于这些担忧和人工智能（AI）的概念交织在了一起。DeepMind并不是普通公司，它是一家专注于人工智能的公司，开发着能够表现出智能推理能力的技术。

去年该公司的工程师们发布了一篇论文说该公司开发出了一个能够复制“专业人类视频游戏测试者”的程序。今年三月份，谷歌DeepMind开发出了一个掌握了有3000年历史的中国棋盘游戏围棋，从而创造了历史；在此之前，因为围棋可能的走子步数的庞大数量，人们认为目前还不具备能掌握围棋的技术。在被视为计算的里程碑的比赛中，该公司的AlphaGo以4-1的成绩击败了世界围棋冠军。

生病前就治好

现在，这样一家公司拥有了包含所有这些人的详细、私密、尽管匿名的医疗记录的数据库，其中包括HIV感染状况、过去的吸毒过量和堕胎信息。DeepMind说其需要这些数据为试图防止急性肾损伤的医院产生医疗报警信号。

一些人担忧的是，DeepMind的数据库可能会被允许用于原本规定之外的目的。公众对NHS病人的隐私没有得到保障的事实已经不陌生了——2014年，因为对医疗机密可能处于风险之中的担忧，政府被迫暂停然后缩减其打造一个单个英文医疗数据库的提案。

DeepMind没有隐藏它与NHS的工作，该公司在二月份宣布其正与该医疗服务合作开发一款被称为Streams的应用，以帮助医生和护士监测肾病患者。它没有公布的是其对包括病人历史病历在内的数据的取用程度。DeepMind得到的不只是几千个患有肾损伤的病人的数据，而是所有这些医院的病人病历。那是数以百万计的机密文件。

该公司表示其需要整个病人数据库才能使Streams有效。这样的数据库的支持者声称有了这样的数据，就能更快地开发出用于诊断疾病的强大软件包。

New Scientist杂志获得了DeepMind和NHS之间的数据共享协议，该协议披露了哪些信息会被提供给DeepMind。这家谷歌公司的技能是在大量数据中识别复杂模式——而NHS是这样的“深度学习”的一座金矿。

这个数据宝库中充满了医院每日的活动日志，例如病人的位置和状态记录——以及谁于何时探望了他们。DeepMind也将获得病人的病理和放射检查记录。

除了实时数据之外，DeepMind还有权使用重症监护和事故与急诊部门的历史记录。理论上，鼓捣这些信息让DeepMind可以基于这些超过任何个人消化和分析能力的数据开发出预测应用。

通过比对患者数据，DeepMind可能能够预测某人处在某种疾病的还没有明显症状的早期阶段。这是医疗领域的圣杯：不是当人们生病时治疗病人，而是在他们生病前就治好他们。

乌托邦？也许吧。这些技术的愿景背后还有我们所生活的这个时代的困境的症结。谷歌、Facebook等公司的生存靠的是这一事实：我们用自己的隐私权换取使用我们的数据开发的新技术。

“智能爆炸”

和苹果一样，谷歌正在打造在医疗应用领域的声誉。而事实上，学术界将机器学习用在医疗上已不是什么新闻了。但是，这些数据的使用权被交给了世界上最大最强的公司之一。这带来了一个问题，即该公司是不是很快就可能在NHS健康分析上变成最大的公司，造成事实上的垄断。

人工智能代表着一些新事物，一个程序能够改进自身并很快超越人类智力的希望。这就是所谓的“智能爆炸”——一个人类迎来自身毁灭的节点。我们离到这一步还有一段距离。即使有这样的目标代价，也还没人开发出尊重社会和道德规范的机器。对人类而言，做到这一点是很困难的。

有些人可能会说这样的推断是荒谬的。毕竟微软的Twitter聊天机器人Tay在变成一个种族主义者和种族灭绝呐喊者之前坚持了几个小时，然后她被关闭了。但正如最初投资过DeepMind的发明者伊隆·马斯克所说，正是对“终结者”技术的担忧驱使他警告其危险。

为了也许听起来商业的原因，DeepMind在人们的注视下进行运作。但这通常带来的问题比给出的答案还多。谷歌在2014年以4亿英镑收购DeepMind时设立的人工智能伦理委员会仍然是技术上最大的谜团之一，这两家公司都拒绝透露谁管理着这一委员会。

人工智能需要数据进行学习，因此谷歌DeepMind要吸取所有这些病历记录。所以为什么要保密？如果病人已被告知了事情的来龙去脉和原因，他们可以做出明智的选择。如果他们认为谷歌在NHS新的关键技术上垄断地位的潜在风险赶不上其所带来的好处，那就让我们辩论吧。但如果该公司不做任何解释，只是秘密进行，公众就肯定不会赞同这样的计划。