NaNoWriMo 的名字听起来就像父母慢慢对婴儿说的话，它是一项年度“挑战”，成千上万看似适应良好的人决定在一个月内写一本小说。举办这项绝妙折磨的非营利组织在其网站上合理地问道： “我需要一些特别的东西来写小说吗？” “不需要！”

全国小说写作月始于 1999 年，当时只有 21 名参与者，如今每年 11 月都有近 50 万人参加。该活动也是将活动游戏化的组织的名称，在其在线平台上接待参与者。要“赢得” NaNoWriMo，你需要在一个月内至少写出 50,000 个单词（大约是《了不起的盖茨比》的长度）——或者每天写出 1,667 个单词，NaNoWriMo告诉我们，“科学家们已经确定这是激发创造力的最佳数量。”

NaNoWriMo 最早出现在旧金山湾区，硅谷的印记无处不在；如果你曾经想过小说创作可以得到优化，那么它就像小说写作界的Soylent 。该组织宣称其平台“像 Fitbit 追踪步数一样为作家追踪单词”。但只要它涉及人类实际坐下来挥洒汗水写出句子，这一切对我这样的人来说似乎都是无害的，我是一个认为写作只是艰苦工作并且不适合所有人的脾气暴躁的人。但在周一，NaNoWriMo 表达了对使用人工智能的看法，事实证明，人类不再是一项要求。

现在我想我知道 NaNoWriMo 的发展方向了，并且我赞同：就让机器人来做吧。

在一份看似可能是由人工智能撰写的声明中，该组织拒绝“明确支持”或“明确谴责”使用技术援助。如果你想反对，NaNoWriMo 辩称，否认人工智能会加剧“阶级歧视和残疾歧视问题”。阶级歧视的论点与“一定程度的特权”可能赋予一些作家“聘请人类进行反馈和审阅的经济能力”有关。残疾歧视的指控更加荒谬。人工智能应该被允许帮助你写小说，因为“并非所有人的大脑都具有相同的能力，并非所有作家都具有相同的教育水平或他们所用语言的熟练程度。”

是的，这就是写作需要努力的原因。如果我参加一场比赛，看看我能否修理一台坏了的洗衣机，由于我缺乏水管工的教育和熟练程度，这会让比赛变得困难，甚至不可能。如果我能访问 YouTube 上的水管工教程视频或使用机器人水管工（如果我们有的话），这项任务就会容易得多。当然，修理洗衣机和写小说是两种不同的成就；自己修理水管可以为你节省几百美元，并可能带来一种满足感，而写小说只会让你自我感觉良好。水管工拥有一项有用的技能，需要通过培训和反复试验获得专业知识，而根据 NaNoWriMo 的说法，其参与者“进入本月时是小学教师、机械师或全职父母。他们离开小说家。”这就是为什么我从来都不喜欢 NaNoWriMo。

网上许多人对该组织的决定感到愤怒，一些作家从作家委员会辞职。人工智能在创意人士中并不受欢迎，即使是兼职创意人士也不受欢迎，因为大型 语言模型已经蚕食了已出版作家的作品，并有可能进一步削弱创造力的价值。许多批评者提到人工智能“窃取”作品的倾向。许多残疾作家尤其对他们应该需要人工智能的想法感到不满。劳拉·艾略特是一位作家，她的处女作将于明年春天出版，她在 X 上写道：“残疾作家不需要不道德的盗窃机器来写作，因为我们缺乏不抄袭的创作能力——鼓励人工智能是对所有作家的侮辱，这种借口是令人震惊的残疾歧视。”

我同情这些作家，他们觉得写作项目背叛了他们，而写作项目显然是他们的一个有益的激励因素。但是，如果不同程度的“教育和熟练程度”将那些能够成功应对挑战的人与那些不能成功的人区分开来，也许每个人都应该再休息一个月。就我个人而言，即使写作很有回报，写作也很难，尽管我已经从事了几十年的写作生涯。随着时间的推移，你会越来越自信，但要让最终写在纸上的东西与你心中的想法相符，总是很困难的。正是这种努力——写作需要的百万种个人选择——赋予了它独特的人性。（也许这么说是亵渎，但也要考虑到，并不是每个人都天生就是作家，或者需要努力成为作家。）

这就是为什么我认为 NaNoWriMo 的声明是个好消息。世界需要更少的小说，当然需要更少的一个月内创作的小说。人工智能渴望分心；我们需要在机器人开始弄乱核密码或社会安全号码之前给它们一些事情做。只需给它们 NaNoWriMo 即可。它们可能可以在几秒钟内写出 50,000 个单词。更好的是，它们还可以阅读其他人工智能创作的小说，让每个人都免于大量糟糕的写作。毕竟，阅读大量材料以将其重构为原创作品是它们最擅长的事情。当人工智能花费数年时间发展自己的能力时——一部接一部地写作和搁置小说——那么也许它们将为我们的人类努力做出贡献。

在此之前，如果你想写，就写吧，但不要以为写出来的东西会很好。也不要以为写出来的东西会很快。当我们只有大脑时，我们需要仔细斟酌每一个字。也许这就是为什么 NaNoWriMo 如此有吸引力，恰恰在这个预测软件盛行的时代：人们想要接受挑战，做一些需要耐心、毅力和想象力的事情，而且会产生不可预测的结果。如果没有这些，你可能就只能摆弄你的 Fitbit 了。

With a name that sounds like something a parent would slowly mouth to their infant, NaNoWriMo is an annual “challenge” in which many thousands of seemingly well-adjusted people decide to write a novel in a month. “Do I need something special to write a novel?” the nonprofit that puts on this exquisite torture reasonably asks on its website. “Nope!”

National Novel Writing Month began in 1999 with 21 participants, and now nearly half a million take part every November. The event is also the name of the organization that gamifies the exercise, hosting participants on its online platform. To “win” NaNoWriMo, you need to produce a minimum of 50,000 words in a month (about the length of The Great Gatsby)—or 1,667 words a day, a number, NaNoWriMo tells us, that “scientists have determined to be the perfect amount to boost your creativity.”

NaNoWriMo first emerged in the San Francisco Bay Area, and it has Silicon Valley’s fingerprints all over it; if you’ve ever thought that producing fiction could be optimized, this is like the Soylent of novel writing. The organization boasts that its platform “tracks words for writers like Fitbit tracks steps.” But as long as it involved humans actually sitting down and sweating out sentences, it all seemed pretty harmless to someone like me, a curmudgeon who thinks writing is just hard work and not for everyone. But on Monday, NaNoWriMo expressed its thoughts on the use of AI, and it turns out that being a human is no longer even a requirement.

And now I think I know where NaNoWriMo is headed, and I approve: Just let the robots do it.

Read: My books were used to train Meta’s generative AI. Good.

In a statement that seemed like it may have been written by AI, the organization refused to “explicitly support” or “explicitly condemn” the use of technological assistance. And in case you thought to object, NaNoWriMo argued that disavowing AI would have exacerbated “classist and ableist issues.” The classism argument had to do with the fact that “a level of privilege” might endow some writers with “the financial ability to engage a human for feedback and review.” The ableism charge was even more absurd. AI should be allowed to help you write your novel because “not all brains have same [sic] abilities and not all writers function at the same level of education or proficiency in the language in which they are writing.”

Well, yes. That’s why writing takes work. If I entered a contest to see if I could fix a broken washing machine, my lack of education and proficiency as a plumber would make that difficult and most likely impossible. Allow me to access YouTube videos of plumbing tutorials or use a robot plumber (if we ever get those), and the task will be much, much easier. Fixing your washing machine and writing a novel are, of course, two different kinds of accomplishments; doing your own plumbing will save you a few hundred dollars and might provide a sense of satisfaction, while the novel will just make you feel good about yourself. Plumbers have a useful skill that demands expertise acquired through training and much trial and error, whereas, according to NaNoWriMo, its participants “enter the month as elementary school teachers, mechanics, or stay-at-home parents. They leave novelists.” This is why I’ve never liked NaNoWriMo.

A lot of people online were angry about the organization’s decision, and a few authors stepped down from its writers’ board. AI is not popular among creative people, even part-time creative people, given that large language models have cannibalized the work of published authors and threaten to further erode the value of creativity. Many of the critics mentioned AI’s penchant for “stealing” writing. A number of disabled writers in particular took offense at the idea that they should need AI. Laura Elliot, an author whose debut novel will be out next spring, wrote on X that “disabled writers do not need the immoral theft machine to write because we lack the ability to be creative without plagiarism—encouraging AI is a slap in the face to all writers and this excuse is appallingly ableist.”

I’m sympathetic to these writers who feel betrayed by a writing project that was apparently a helpful motivator for them. But if varying levels of “education and proficiency” divide those who can succeed at the challenge from those who can’t, maybe everyone should just take another month. Personally speaking, writing is difficult even when it’s rewarding, even after I’ve spent a decades-long career doing it. You gain confidence over time, but it’s always a struggle to make what ends up on the page correspond with what was in your mind. That struggle—the million individual choices that writing demands—is what gives it its particular human flavor. (And maybe it’s sacrilege to say this, but consider, too, that not everyone was born to be an author or needs to try to become one.)

Read: Murdered by my replica?

Which is why I, for one, think that NaNoWriMo’s statement is great news. The world needs fewer novels, certainly fewer novels that have been written in a month. And artificial intelligence is itchy for distractions; we need to give the robots something to do before they start messing with nuclear codes or Social Security numbers. Just give NaNoWriMo to them. They can probably produce 50,000 words in a few seconds. Better yet, they can also read the novels that other AIs produce, saving everyone from a lot of bad writing. Reading metric tons of material in order to reconstitute it as original work is, after all, what they do best. When the AIs have spent years developing their abilities—writing and shelving novel after novel—then maybe they will have something to contribute to our human efforts.

Until then, if you want to write, just write, though don’t assume it will be good. And don’t assume it will be quick. When all we have is our human brains, we need to deliberate on every word. Maybe that’s why NaNoWriMo has had such appeal, precisely in a time of prediction software: People want the challenge of doing something that requires patience and persistence and imagination, and that spits out unpredictable results. Take that away, and you might as well just be fiddling with your Fitbit.