We’ve all been there. You spend no more than five minutes talking to your friend about how you need new workout shoes, and next thing you know, hundreds of Nike ads show up in your Facebook feed. Coincidence? — It’s not.
我们都去过那里。 您只花了不到五分钟的时间就与您的朋友谈论如何需要新的健身鞋，接下来您就会知道，Facebook订阅源中会显示数百个耐克广告。 巧合？ — 不是 。
But the tech giants aren’t directly listening to your conversations to get data and feed you specific ads, they do it in a more subtle and cheaper way.
证据怎么说。 (What the evidence says.)
Despite the endless conspiracy theories, there isn’t any actual evidence that Google, Facebook, or any other major tech company is actively recording user voice data without their knowing.
The Cambridge Analytica scandal gave us with an idea of the nightmare that would face a company caught secretly recording and sharing private user information.
Of course this doesn’t mean that everybody learned their lesson and that won’t happen again, but it puts companies who want to do it at a huge risk just to have a little more user data. Truth be told, we already give them so much information for free and without even realizing it.
While it is true that smartphones that have a “Hey Siri” or “Okay Google” function are constantly listening, they are not actively processing the information they hear. Keyword triggers like “Hey Siri” need to be used in order to wake up the device from a low power state to a more powerful listening state, in which then it starts recording.
具有“ Hey Siri ”或“ Okay Google ”功能的智能手机确实在不断监听 ，但它们并没有积极地处理所听到的信息。 为了将设备从低功耗状态唤醒到更强大的聆听状态，需要使用诸如“ Hey Siri ”之类的关键字触发器，然后才能开始记录。
语音识别软件比您想象的更昂贵，更复杂 (Voice recognition software is more expensive and complex than you think)
If you’re not convinced yet, let’s make some estimations. If smartphones were to be listening to you 24/7, manufacturers would have to increase the number of keywords to thousands more — as all topics would have to be covered. That requires more processing power, and your battery would drain really fast, which would be quite noticeable.
如果您还不确定，让我们做一些估算。 如果智能手机要全天候24/7收听您的声音，那么制造商必须将关键字数量增加到数千个，因为必须涵盖所有主题。 这需要更多的处理能力，并且您的电池会很快耗尽，这非常明显。
Moreover, keyword detection doesn’t provide context. Imagine you’re complimenting a friend on his new pair of sunglasses, that doesn’t mean you necessarily want to buy a new pair of sunglasses; therefore, that information is extremely irrelevant.
而且，关键字检测不提供上下文。 想象一下，您在称赞一位朋友的新太阳镜，这并不意味着您一定要买一副新太阳镜。 因此，该信息是完全无关的。
A good voice recognition system would have to process and analyze all of your conversations to pick out relevant keywords and sentences that are in context with the desire of new products, places, and other categories for advertisers to use.
一个好的语音识别系统必须处理和分析您的所有对话，以挑选出与广告客户想要使用的新产品 ， 新地点和其他类别相关的相关关键字和句子。
But let’s assume smartphones can send all voice data to external servers for further processing. If that were the case, each smartphone would collect about 59GB of data a year per user for 24/7 monitoring. And even cutting that down from 24-hour monitoring to just one hour a day — which is certainly not realistic — requires 2.5GB per user a year. That means that we would need approximately 6 Exabytes for the 2.5 billion smartphone users in the world — for just one hour a day. That’s a huge amount of data to process and store.
And that’s without taking into account the cost and time of processing all that voice data. Google sells its Speech Recognition system for $0.006 per 15 seconds of audio. So recording one person only one hour a day would cost $525 a year. If we scale that up to the 2.5 billion smartphone users in the world, that’s $1.31 trillion just for voice processing. And that’s without the cost of data storage, processing the transcripts, and other expenses.
而且这没有考虑处理所有语音数据的成本和时间。 Google出售其语音识别系统 ，每15秒音频$ 0.006。 因此，一天只录制一个人一个小时，一年就要花525美元。 如果我们将其扩展到全球25亿智能手机用户，那么仅语音处理就达到了1.31万亿美元。 而且这没有数据存储，处理成绩单和其他费用的成本。
Processing everyone’s voice for just 1 hour per day would consume the entire 2018 digital ad budget.
It is also relevant to keep in mind that language plays a big role in all of this. For instance, I’m from Barcelona — Spain — so I speak Catalan with my family and friends, but my phone is set in English. If my smartphone had to actually process the voice inputs it receives, it would have to recognize another language and process that information differently. That would require really complex software and additional costs.
同样重要的是要记住，语言在所有这些方面都起着重要作用。 例如，我来自西班牙的巴塞罗那，所以我与家人和朋友说加泰罗尼亚语，但是我的手机设置为英语。 如果我的智能手机必须实际处理它收到的语音输入，则它必须识别另一种语言并以不同方式处理该信息。 这将需要非常复杂的软件和额外费用。
进行测试 (Put it to the test)
Here are two articles that tested the theory that smartphones listen to us.
您的智能手机实际上是如何“监视”您的。 (How your smartphone actually ‘spies’ on you.)
So up until this point, we’ve established that smartphones don’t actively listen to us and sell that data to advertising companies. So here is the main reason why we get ads that make us think otherwise.
大数据 (Big data)
This is the cheapest and most efficient method of data acquisition, and it has been used for years. Big data collection allows companies to learn a lot about us by just pulling data from different sources.
Truth is, advertising firms don’t need to hear our discussions to accurately define a consumer profile. Location data, browsing history and behaviour, IPs, social media profiles, website cookies, spending habits, etc, all provide enough information to predict what you might be thinking of purchasing.
Big data can be so accurate as to anticipate our wants before we even realize them.
Google, Facebook, Amazon… all the tech giants have so much information about you — that you agreed to give them for free — that is sold to advertising companies. Those firms then sort us into categories based on demographics, interests, and personal relationships; and companies that are interested in that specific group of people pay to pitch advertisements to.
Google，Facebook，Amazon…所有科技巨头都拥有关于您的大量信息，您同意免费提供给您，这些信息已出售给广告公司。 然后，这些公司根据人口统计，兴趣和个人关系将我们分类。 而对特定人群感兴趣的公司则向其投放广告。
With Big Data and by connecting the dots on the digital trail you leave, advertising companies are able to perfectly target you and those you interact with — remember, everything is connected.
大数据滥用 (Misuse of Big Data)
Big Data is extremely insightful, and the tech giants could misuse that information by selling it to the wrong people or by using it to bias citizens, like what happened with the Cambridge Analytica scandal.
In this day and age, data is power.
Once you’ve downloaded and authorized an app — which you need to do in order to use it — its creators have access to whatever data they want on your device. Deleting it afterwards does nothing to protect your privacy, as your data has been sent and stored for future use. Even if one of your friends authorized an app on their phone, that app still manages to collect a lot of your data. And with the hundreds of data points we are constantly giving out, this can be scaled to impact people of certain vulnerable demographics for not-so-ethical things.
So no, your phone is not listening to you 24/7 to sell your conversations to third parties; it simply isn’t possible technologically or economically. This theory still prevails because people aren’t aware of the power that Big Data holds.
所以不， 您的电话没有全天候24/7听您向第三方出售您的对话； 在技术或经济上根本不可能。 该理论仍然盛行，因为人们并不了解大数据所具有的力量。
Data tracking is very real and highly invasive, and we should all be concerned about our privacy, especially in light of recent data leaks and shady deals. So don’t worry about your smartphone listening to you — or at least not yet.