074:关于「人工智能替代人类」,故事的完整版本是这样的

AlphaGo 本周与柯洁比赛在有关方面的干预下被大大降低了新闻报道的权重,你可以将其理解为有司担心人工智能胜利后引发民众恐慌,你可以认为这是为了防范公众过度讨论 Google 的一种举措,但无论如何,正如我多次在会员通讯里的判断:这次比赛之于围棋、之于人类,其意义已微乎其微。

考虑到围棋在西方世界是一个小众游戏,这几天美国媒体对比赛的报道也仅仅局限在 AlphaGo 又一次击败了人类围棋选手层面,除此之外并没有太多深层次的讨论。反而是 Steven Levy 对于 20 年前「深蓝」打败卡斯帕罗夫的回忆更有共鸣,当时 Levy 还是《商业周刊》的记者,他回忆了报道出炉时的情形:

In my own tribe’s form of jousting, I had campaigned for the cover, despite the editor’s declaration that “we will never run a cover about chess.” I successfully argued that this was not about a game of chess, but rather about a much more epic contest between human and artificial intelligence. What clinched it was the cover line I suggested: “The Brain’s Last Stand.”

img

事实上,「深蓝」出现的 1997 年还处在人工智能又一个「冬天」里,全世界也还在 PC 互联网即将爆发的前夜。如今,人工智能再次得以复兴,世界也从 PC 互联网快速演进到移动互联网,并正在大踏步来到后移动互联网时代。Levy 不无感慨地写道:

Amazingly, when the Deep Blue match occurred, AI was in its “winter” peri0d. Now it is flowering. We hear of amazing machine learning accomplishments on a daily basis. But in 2017, we view them differently. We view them as inevitabilities.

The prime example is last year’s contest, during which DeepMind’s AlphaGo program thumped an 18-time world champion in a series of five games. Go is a much more challenging feat for a computer than chess. Yet AlphaGo did not need to resort to any of the tactics that IBM used to distract, deceive, and ultimately destroy Kasparov. The human champion, Lee Sedol, ended with respect for his opponent and awe for how far computer science had come. But though the match deservedly received attention, it was nowhere near as mythic as the Deep Blue match was. The ground has shifted. Given enough time, money, and machine learning, there’s no cognitive obstacle that machines will not surmount.

对此,Levy 担心地认为,或许我们更应该担心研发这些机器或部署这些机器的人。尽管 Levy 马上承认自己有点情绪化,毕竟作为报道硅谷技术发展长达 30 年的老记者,Levy 很清楚当下人工智能与世人期望之间的巨大差距,比如同样是在本周发生的一件事,一直以来以人工智能作为自己「员工」的 Facebook,居然还有 4500 人的内容审核团队……

英国《卫报》获取了一份据说是 Facebook 内容审核手册的文件,其中有几个细节:

  • Facebook 的全球审核员工多达 4500 人;
  • 这些审核员工具有决定某些内容是否需要删除的权限;
  • 威胁杀死特朗普的帖子会被删除,更近一步,威胁各国政府首脑的帖子都会被处理;
  • 普通人之间的死亡、暴力威胁则被保留;
  • 允许用户直播自我伤害,比如自杀;

img

这个手册曝光引发了诸多讨论,从社会伦理层面上说,Facebook 对于内容是否删除的界定是否符合道德、法律的争议一直存在,Facebook 官方在《卫报》报道后公开回应:内容审核却是会有所谓的「灰色区域」,比如讽刺幽默和不雅内容之间的界线并非十分清楚。而根据 FT 的报道,英国政府对于 Facebook 的内容审核政策有诸多不满,FT 援引一份来自文化、媒体和体育事务特别委员会的报道称:

可耻的是,最大、最富有的社交媒体公司远未采取足够行动来处理非法或危险内容。

而站在互联网的层面来说,Facebook 还拥有如此庞大内容审核队伍的消息依然令人震惊。今年 3月,FB 计划上线一个能够自动判断用户是否有自杀或自残倾向的人工智能系统,这个系统通过对用户长期起来发布内容的倾向进行判断,如果系统认为某个内容展示了发布者的自杀倾向,就会提示给其他阅读该内容的用户,让他们选择是否将这个内容反馈给 FB 的内容审核人员。

不过这个系统并不能「自主决策」,如果得不到其他用户的支持,上述提示不产生任何决定性的作用。更重要的是,当 FB 一面宣传自己的算法可以帮助用户搜索到照片里的特定人或物的时候,一面却在应对诸如照片里的情色信息或涉及儿童色情方面困难重重。在《卫报》得到的手册里,有以下一组数据:

  • Facebook 每个月要处理近 54000 份复仇色情和性爱勒索事件;
  • 今年一月,Facebook 向员工透露,为处理这类性虐待,他们每个月要禁用 14000 个账户,而在这些事件中,33%都涉及儿童;

这当然无法证明 FB 的人工智能就一定碌碌无为,但从一个侧面显示了人工智能在当下面临的尴尬境地——在大公司一波又一波声势浩大的宣传里,在投资人、创业者的豪言壮语里,人工智能仿佛一夜之间可以改变人类生活、工作以及现实与虚拟的全貌,可另一面则是这些公司「偷偷摸摸」地使用人类完成本该由机器完成的工作。

无独有偶,今年 4 月份的《连线》杂志也报道了一个类似的案例,这个案例的主角是 Google。他们雇佣人类作为机器的「教练」,但是这些人类教练并不是高大上的科学家或工程师,而是一些没有多少学历的低端劳动力,他们中的绝大多数人都是临时工,他们的工作情况大概就是这样的:

Tech companies have long employed content moderators; as people upload and share more and more content, this work has become increasingly important to these internet giants. The ad raters WIRED spoke with explained that their role goes beyond monitoring videos. They read comment sections to flag abusive banter between users. They check all kinds of websites served by Google’s ad network to ensure they meet the company’s standards of quality. They classify sites by category, such as retail or news, and click links in ads to see if they work. And, as their name suggests, they rate the quality of ads themselves.

而在 3 月份,Google 的 Youtube 上线了新的广告系统,也给这些人类教练带来新的挑战:

……This new project meant focusing almost exclusively on YouTube—checking the content of videos or entire channels against a list of things that advertisers find objectionable. “It’s been a huge change,” says one ad rater.

Raters say their workload suggests that volume and speed are more of a priority than accuracy. In some cases, they’re asked to review hours-long videos in less than two minutes. On anonymous online forums, raters swap time-saving techniques—for instance, looking up rap video lyrics to scan quickly for profanity, or skipping through a clip in 10-second chunks instead of watching the entire thing. A timer keeps track of how long they spend on each video, and while it is only a suggested deadline, raters say it adds a layer of pressure. “I’m worried if I take too long on too many videos in a row I’ll get fired,” one rater tells WIRED.

FB、Google 背地里的这些动作与他们各自在媒体上制造的那种技术至上的形象形成了一个鲜明的对比,也有助于让我们更好地认清楚所谓机器取代人的荒谬,我曾在第 68 期会员通讯里详细讨论过「此人工智能与彼人工智能」,其关键就在于,没有哪家公司在开发可能超越人类的超级智能,而是以「人工智能」包装自己过往的产品,如私有云(亚马逊、阿里云)、如数据库(Oracle)、如企业软件(Selasforce)、如摄像头(苹果的 iPhone)等等,而恰好,深度学习在语音、图像上的突破,为这些大公司的某些产品带来新的发展机遇,这才是故事的完整版本。

Zhao Saipo

View posts by Zhao Saipo
赵赛坡,科技博客作者、资深科技观察家、付费科技评论 Dailyio 创始人、出品人,覆盖 3000+ 付费用户。 曾担任 TechTarget 中国区记者、频道主编、AI 自媒体「机器之心」前联合创始人。

发表回复