and how humans lost.

以及人类如何迷失。

None of this is real. It’s 21st century disinformation warfare fought on the battlefield of your timeline.

这些都不是真实的。 在您时间轴的战场上进行了21世纪的虚假信息战。

Over past months I’ve studied how weaponised AI bots on social media create chaos and division within our society. This anticipatory research has informed some startling conclusions – the implications of which will shock you.

在过去的几个月中，我研究了社交媒体上武器化的AI机器人如何在我们的社会中造成混乱和分裂 。 这项前瞻性研究已经为您提供了一些惊人的结论，其中的含义将使您震惊。

For research, I ran advertising campaigns on Facebook, Twitter and Instagram to study who or what was engaging. I got some big hitters on social media to engage with me publicly on Twitter; @DrTedros, @PiersMorgan and @Afneil, who have a combined total of 10 million followers.

为了进行研究，我在Facebook，Twitter和Instagram上进行了广告活动，以研究谁参与了。 我在社交媒体上吸引了一些大手笔，可以在Twitter上与我进行公开互动。 @DrTedros ， @PiersMorgan和@Afneil ，总共有1000万关注者。

This brought swarms of attack bots to my timeline along with some very nasty people. At least I think they were people. It’s hard to tell, as you’ll soon find out. It also resulted in my Facebook and Instagram accounts getting hacked and permanently disabled.

这将攻击机器人群和一些非常讨厌的人带到了我的时间表。 至少我认为他们是人。 这很难说，因为您很快就会发现。 这也导致我的Facebook和Instagram帐户被黑客入侵并永久禁用。

Fifteen years ago, give or take, you could buy fake followers for your social media accounts. Generated in their millions by automated scripts designed by code monkeys making easy money behind a website. You pay the money and the code tells these newly acquired followers who to follow, what to like, retweet, view and, well, you get the picture. The whole thing was automated from the moment you paid.

十五年前，无论付出还是付出，您都可以为自己的社交媒体帐户购买假粉丝。 由代码猴子设计的自动脚本以数百万美元的价格产生，这些猴子在网站背后赚钱容易。 您付了钱，代码告诉了这些新获得的关注者谁该关注，喜欢什么，转发，查看，以及您得到的图片。 从您付款的那一刻起，整个过程就自动化了。

The bot industry was born.

机器人产业诞生了。

These bots were dumb — they didn’t do a whole lot, they followed, they liked, they wrote canned responses on Twitter, YouTube and Facebook. Their ease of detection was a problem however, leading to numerous “influencers” being called out and embarrassed. It didn’t take long for the mischievous to buy bots for someone else to discredit them later by calling them out.

这些机器人很笨-他们没有做很多事情，他们遵循，喜欢，他们在Twitter， YouTube和Facebook上写了罐头回复。 但是，它们易于检测是一个问题，导致许多“影响者”被召唤出来并感到尴尬。 淘气者花了很长时间才购买了机器人，让其他人稍后通过叫出它们来抹黑它们。

These were Bots 1.0 if you like. Their job was to make you seem important and worth following because how could 50,000 followers/views/likes be wrong?

如果您愿意，这些是Bots 1.0。 他们的工作是使您看起来很重要并且值得关注，因为50,000个关注者/观点/喜欢怎么可能错了？

To keep the money rolling the smarter code warriors improved their bot craft by making them harder to detect. These new bots had more realistic profiles with pictures stolen from the internet plus new behaviours such as following and interacting with each other. Better still, these bots would create the illusion of a human operator by posting their own material and conversing with humans or other bots. After all, natural language processing has been with us since the 1960s and ELIZA. Some of these bots were so convincing that in a reversal, humans would follow bots without knowing.

为了保持资金流向，更聪明的代码战士通过使其更难被发现来改善了他们的机器人技术。 这些新的bot具有更逼真的配置文件，其中包括从互联网上窃取的图片以及新的行为，例如互相关注和互动。 更好的是，这些机器人会发布自己的资料并与人类或其他机器人进行对话，从而给人类操作员带来错觉。 毕竟，自1960年代和ELIZA以来，自然语言处理一直伴随着我们。 这些机器人有些令人信服，以至于人类在逆转时会不知不觉地跟随机器人。

Let’s call these Bots 2.0. You can still buy them. Their job is to amplify your message through retweets, reposting, natural language interactions that encourage humans to do the same. It stands to reason that if so many people were engaging it must be true.

我们将其称为Bots 2.0。 您仍然可以购买它们 。 他们的工作是通过转推，重新发布自然语言交互来鼓励您这样做，从而扩大您的信息。 有理由认为，如果有那么多人参与，那一定是真的。

Today, the bot business model has changed. The real action has moved from entrepreneurial coders selling bots to narcissists, to government agencies and other actors vacuuming up the best tech talent. We’re talking about teams with the best hackers, coders, data scientists, digital biologists, behavioural experts and machine learning gurus money can buy.

如今，机器人商业模式已经改变。 真正的行动已经从销售机器人的企业家编码员转变为自恋者，再到政府机构和其他演员，他们吸纳了最好的技术人才。 我们正在谈论的团队可以买到最好的黑客，编码人员，数据科学家，数字生物学家，行为专家和机器学习专家。

The bots coming out of these factories are smart; AI enabled, informed by machine learning with profile pictures created with deep fake technology. If you watch carefully you can still spot them. For example, if you overlay a series of profile pictures you’ll find that miraculously their eyes all line up. If your mind is attuned you’ll spot patterns, repeated or derivative comments. But it’s getting harder and soon you will never know.

从这些工厂出来的机器人很聪明。 启用AI并通过机器学习提供信息，并使用深度伪造技术创建的个人资料图片。 如果您仔细观察，仍然可以发现它们。 例如，如果叠加一系列个人资料图片，您会发现他们的眼睛都奇迹般地排列在一起。 如果您的想法很协调，您会发现模式，重复的或衍生的评论。 但是它变得越来越困难，很快您将永远不知道。

Now armed with weapons grade computing power these bots are getting smarter because they feed on our discourse and play us back to ourselves. We are literally feeding the trolls, the more we play the better the AI. What we’re seeing are heat-seeking bot swarms, like bees on the wing, designed to be attracted to the noise signature generated by human actors. These swarms move around social media to amplify and legitimise specific messages, statements and ideas that alter our perception of reality.

现在，这些机器人拥有武器等级的计算能力，它们变得越来越聪明，因为它们以我们的话语为食，并把我们玩回了自己。 我们实际上是在喂巨魔，我们玩的越多，人工智能就越好。 我们所看到的是热寻机器人群，就像机翼上的蜜蜂，旨在吸引人类演员产生的噪音信号。 这些群体在社交媒体上四处走动，以放大和合法化改变我们对现实的看法的特定消息，陈述和想法。

One can post the most odious statements in a public forum that only 20 years ago wouldn’t have been supported and perhaps they still wouldn’t be if not for the amplification by bots. Social norms and behaviour have been cast aside as if they are relics from a bygone age. These amplified statements serve to embolden, unite and legitimise the beliefs of people who had hitherto been banished to the fringes of society. Heck, you don’t even need a human to kick it off. If they’re unknowingly following bots that insert messages into their echo chamber they may very well pass it on before its legitimised by known person.

人们可以在一个公共论坛上发表最可恶的声明，直到20年前它才得到支持，如果不是因为机器人的扩大，也许它们仍然不会得到支持。 社会规范和行为已被抛弃，就好像它们是过去时代的遗物一样。 这些夸大的言论有助于使迄今被流放到社会边缘的人们的信念变得胆大，团结和合法化。 哎呀，您甚至不需要人类来启动它。 如果他们在不知不觉中跟随将消息插入回声室的僵尸程序，他们很可能会在已知人将其合法化之前将其传递出去。

Perhaps someone with 85.5 million followers.

也许有人拥有8550万追随者 。

These are bots 3.0. Their job is to seed chaos and division with swarms backing both sides of every argument.

这些是机器人3.0。 他们的工作是在每个论点的两面都支持一群混乱和分裂的种子。

There is no grey, there’s just black or white. There is no nuance, you’re with us or against us, yes or no, who needs bombs when you’ll do it yourself?

没有灰色，只有黑色或白色。 没有细微差别，您支持我们还是反对我们，是或否，您自己动手时谁需要炸弹？

To quote The Joker,

为了引用小丑 ，

“As you know, madness is like gravity…all it takes is a little push.”

“如您所知，疯狂就像万有引力……所需要的只是一点推动力。”

This manipulation of public thought via the media isn’t new. What is new is the scale of operation and the symbiosis between social media platforms, personalities and bots.

通过媒体操纵公众思想并不是什么新鲜事。 新功能是运营规模以及社交媒体平台，个性和机器人之间的共生关系。

Platforms have no motivation to root out these increasingly undetectable bots. Outrage is an easier sell to keep their platforms sticky, attract new users and keep them engaged. More users, better share price as these bots provide the kind of engagements that advertisers pay for. So what if it’s a lie.

平台没有动机来根除这些日益难以发现的机器人。 愤怒是一种易于出售的方法，可以保持其平台粘性，吸引新用户并保持参与度。 这些机器人提供了广告客户付费的互动方式，从而吸引了更多用户，更好的股价。 那如果是谎言怎么办。

The vanity of social media personalities knows no bounds. They are never going to complain about more followers, engaging with and amplifying their thought of the moment. Besides if you’re an influencer, you’re getting paid. Often those who post the most vile statements, emboldened by bots, are rewarded with the most followers. Ker-ching! Everyone wins.

社交媒体人物的虚荣心无止境。 他们永远不会抱怨更多的追随者，参与并扩大他们对这一刻的想法。 此外，如果您是有影响力的人，您将获得报酬。 通常，那些在机器人的鼓舞下发表最卑鄙的言论的人会得到最多关注者的奖励。 柯清！ 每个人都赢。

Except. They. Don’t.

除。 他们。 别。

This vicious circle permits the bots to thrive even as society burns, the flames are fed with our inexhaustible supply of anger.

即使社会在燃烧，这种恶性循环也使机器人得以壮成长，我们的取之不尽用之不竭的怒火已注满了火焰。

An analysis of the followers of some personalities on, for example, Twitter suggests that up to one third are fakes or bots. The personality might not even know as they bask in their quantifiable popularity. Could it be that of the 300 million Twitter users, 100 million are bots?

对Twitter等某些个性的追随者进行的分析表明，多达三分之一的人是假货或机器人。 个性也许甚至不知道，因为他们沐浴在可量化的知名度中。 难道在3亿个Twitter用户中，有1亿是机器人 ？

Who knows?

谁知道？

It’s the like the emperor’s new clothes, we know the bots exist but there are too many vested interests in maintaining the lie to point them out.

就像皇帝的新衣服一样，我们知道机器人的存在，但是保持谎言指出它们有太多的既得利益。

In a game like this, the only way to win is not to play.

在这样的游戏中，唯一的获胜方法就是不玩。

Whilst researching for this article my Facebook account was successfully hacked twice by unknown actors. My Facebook and Instagram accounts have now been permanently disabled.

在研究本文的同时，我的Facebook帐户被不知名的参与者成功入侵了两次 。 我的Facebook和Instagram帐户现已被永久禁用。