08.13.23: Inner Coach
本周的主题的是 Inner Coach 内在教练。
它来自于 Scott Belsky 的一篇短文:
I’ll often whisper to myself towards the end of a hard run, “Come on, give me another sprint…another mile, you can rest later.” This “inner coach” voice has developed over recent years. It celebrates a good run and calls me out when I’m sloppy. When running around the Central Park reservoir, there is a certain stretch I call “the straight shot.” When I reach this point about two miles into my morning run when I am in NYC, my inner coach reminds me to sprint it from start to finish. Doesn’t matter if I’m on a recovery run, feeling under the weather, or exhausted. I do it to remind myself that I always can despite the circumstances, and my inner coach won’t let me get away with an excuse. I’ve come to realize this inner voice is what keeps us honest with ourselves and performant across many parts of our life. All too often we think that, if nobody else notices a short-cut we take, that we can get away with it. But your inner coach cannot be fooled. Your inner coach stubbornly remembers everything. And the self-reliance, strength, and commitment to raw truth that result from empowering and respecting your inner coach makes you better. Develop this inner coach persona in your head. And pay attention to your terrain for anything that can be merchandised to yourself (and team) as a moment of challenge like my “straight shot” around the reservoir and tuck these sprints in.
在一次艰苦的跑步结束时,我经常会低声对自己说:「来吧,再冲刺一次……再跑一英里,你可以稍后休息。」这种「内部教练」的声音是近年来发展起来的。它会庆祝我跑得好,并在我马虎时叫我出去。当围绕中央公园水库跑步时,有一段我称之为「直线射击」的距离。当我在纽约晨跑大约两英里时到达这一点时,我内心的教练提醒我从头到尾冲刺。无论我是在恢复跑步、感觉不舒服还是精疲力竭,都没关系。我这样做是为了提醒自己,无论情况如何,我总是可以做到,而且我的内心教练不会让我以借口逃脱。我逐渐意识到,正是这种内心的声音让我们对自己诚实,并在生活的许多方面表现出色。我们常常认为,如果没有其他人注意到我们走了捷径,我们就可以逃脱惩罚。但你的内在教练不会被愚弄。你的内在教练顽固地记住了一切。授权和尊重你的内在教练所带来的自力更生、力量和对原始真理的承诺会让你变得更好。在你的头脑中培养这种内在的教练形象。并注意你的地形,寻找任何可以向你自己(和团队)推销的挑战时刻,比如我在水库周围的「直线射击」,并将这些冲刺塞进去。
原文讲了 12 条关于跑步的思考,这是其中一条。说起跑步,算是今年我重新拾起来的一个习惯,从春天开始,到现在已经跑了小半年,身体状态在日复一日的打磨中逐渐变好,看到这篇文章也会更感同身受。
而 Inner Coach 的这个想法却是新的。它是一种内在的精神力量,构建在人对自身力量的想象之上,在无时不刻的自我提示中,完成新的挑战。自从读完了这段话,我开始尝试延长跑步的距离,竟然每一次都成功了,虽然只是从 3 公里延长到 5 公里——相比于朋友圈里面的轻松晒出来的成绩相形见绌,但也是不小的进步。之前,到了接近 3 公里距离的时候,我就会开始想象停下来放松的走路,以及回家在沙发上喝冷饮的情景;而现在,Inner Coach 替代了这种想象,他在内心呼唤:无论如何都要继续跑下去。
这是自我安慰吗?或许是的,但它的确带来了改变。我倾向认为,它或许并没有提高我的体能,但它让我把体能更完全的释放出来了——但我还没有尝试过用它来突破身体的极限,比如跑一个前所未有的长距离。
这就把我们带向了 Sam Lessin,在接受 The Generalist 的采访时,他这样讲道:
What trait do you value most highly in others?
I help keep myself honest by soliciting feedback. I am very into sharing shitty ideas at high velocity and relying on others to keep me intellectually in check. I think that’s a mental position worth cultivating: one where you’re willing to be wrong on almost anything, all the time, without stressing it. Trying to avoid being wrong is a path to being intellectually dishonest with yourself and others.
你最看重别人的什么特质?
我通过征求反馈来帮助自己保持诚实。我非常喜欢快速分享糟糕的想法,并依靠别人来控制我的智力。我认为这是一种值得培养的心理状态:你愿意在几乎所有事情上一直犯错,而不用强调它。试图避免犯错是在理智上对自己和他人不诚实的一条途径。
Sam Lessin 的这个观点在很多投资人身上(或者嘴上)并不少见。我之前分享过的 Strong Opinions, Weakly Held 其实也是类似的意思。这个观念最早来自于未来学家、Stanford 教授 Paul Saffo,他这样写道(抱歉,因为原文很短,所以我几乎引用了其中的一半):
The point of forecasting is not to attempt illusory certainty, but to identify the full range of possible outcomes. Try as one might, when one looks into the future, there is no such thing as “complete” information, much less a “complete” forecast. As a consequence, I have found that the fastest way to an effective forecast is often through a sequence of lousy forecasts. Instead of withholding judgment until an exhaustive search for data is complete, I will force myself to make a tentative forecast based on the information available, and then systematically tear it apart, using the insights gained to guide my search for further indicators and information. Iterate the process a few times, and it is surprising how quickly one can get to a useful forecast.
Since the mid-1980s, my mantra for this process is “strong opinions, weakly held.” Allow your intuition to guide you to a conclusion, no matter how imperfect — this is the “strong opinion” part. Then –and this is the “weakly held” part– prove yourself wrong. Engage in creative doubt. Look for information that doesn’t fit, or indicators that pointing in an entirely different direction. Eventually your intuition will kick in and a new hypothesis will emerge out of the rubble, ready to be ruthlessly torn apart once again. You will be surprised by how quickly the sequence of faulty forecasts will deliver you to a useful result.
预测的目的不是尝试虚幻的确定性,而是确定所有可能的结果。无论人们如何努力,当人们展望未来时,都不存在「完整」的信息,更不用说「完整」的预测了。因此,我发现实现有效预测的最快方法通常是通过一系列糟糕的预测。我不会在完成对数据的详尽搜索之前保留判断,而是强迫自己根据现有信息做出初步预测,然后系统地将其分解,利用获得的见解来指导我搜索进一步的指标和信息。重复这个过程几次,令人惊讶的是人们能够如此快速地获得有用的预测。
自 20 世纪 80 年代中期以来,我对这一过程的口头禅是「强观点,弱坚持」。让你的直觉引导你得出结论,无论多么不完美——这就是「强观点」部分。然后——这就是「弱坚持」部分——证明自己是错的。进行创造性的怀疑。寻找不合适的信息,或者指向完全不同方向的指标。最终你的直觉会发挥作用,一个新的假设将从废墟中出现,准备再次被无情地撕碎。您会惊讶地发现,一系列错误的预测竟然能够如此迅速地为您提供有用的结果。
「强观点,弱坚持」和 Sam Lessin 的「通过反馈来保持诚实」是通过外部反馈来矫正自己的判断,而 Scott Belsky 的 Inner Coach 则是通过内在的精神力量来实现自我推动,看起来它们是截然不同的力量源泉。但只要深入想一层,就会意识到,Inner Coach 最重要的作用是强迫自己打开自己,迎接而不是对抗外部挑战,这就意味着,你可以跑更长的距离,或者倾听更强烈反对的意见,并且能够在外部挑战进入之后,仍然能够正常处理这些信息(熵增),并把它们最终消化吸收为内在秩序的一部分。
另外一点观察是:因为 AI 的出现,以及它所带来的存在主义危机,迫使人类社会开始重新思考一些更基本的概念,比如:学习和智能的本质。这些概念太过于稀松平常,在没有经历外部挑战的时候,大部分人不会关注这些名词——除非他们被孩子追问。
机器和孩子都可以被认为是一种外部挑战,就像我在跑到 3 公里的时候急促的呼吸一样,而显然机器现在带来了更大的挑战。在宏观意义上,一些人陷入了更深重的担忧,对于意义、社会、经济等等议题都再次找到了失去已久的媒体版面;在微观层面上,更多人开始不自觉的产生行动,唾手可及便利性是最好的轻推(nudge),人们忍不住把手伸向更强大的机器,只为此刻的满足,而不去考虑长期和整体。
Jerry Neumann 发表了一篇题为 AI and the Structure of Reasoning 的长文,通过对 AI 发展历史的回顾,讨论了当代 AI 技术的局限性和潜力。其主要观点是:虽然有一天可能会出现超越人类想象的人工智能,但目前的生成式人工智能技术并不比人类聪明,也无法自我改进以达到人类的想象。人工智能的第一波浪潮主要是演绎推理,而第二波浪潮则侧重于归纳推理。归纳法采用已知事实并创建规则来推断新事实,比演绎法更普遍,演绎法使用已知规则和现有事实创建新事实,被认为是人类智能的标志。然而,要实现通用人工智能(AGI),需要开发一种超越归纳的新型智能。
他写道:
But if you can’t deduce induction and you can’t induce induction, why do we believe in induction? Just as deduction is believable because of induction, there must be a more general form of reasoning that allows us to use induction comfortably. I’ll call it Invention for the moment, but it is clearly something humans can do that encompasses induction (and thus also encompasses deduction.) There must be layers of inference, from the most specific to the most general. The most specific, deduction, offers the most assurance of truth, and that assurance wanes as inference becomes more general. Invention might not even be inference at all. Regardless, it is a way to think, and so a type of intelligence.
但如果你不能演绎归纳法,也不能归纳归纳法,那我们为什么还要相信归纳法呢?正如演绎因归纳而可信一样,必须有一种更一般的推理形式让我们能够轻松地使用归纳。我暂时将其称为发明,但这显然是人类可以做的事情,它包含归纳(因此也包含演绎)。必须有多层推理,从最具体到最一般。最具体的演绎提供了最真实的保证,而随着推论变得更加普遍,这种保证就会减弱。发明甚至可能根本不是推理。无论如何,它是一种思考方式,也是一种智力。

作者认为,当代以大模型为基础范式的 AI 可以认为是归纳式 AI(induction),本质是基于过去的大量数据输入来推测未来事件发生的概率。他引用 Sam Altman 的话说:
Sam Altman, CEO of OpenAI said “I think we’re at the end of the era where it’s going to be these, like, giant, giant models. We’ll make them better in other ways.”
OpenAI 首席执行官萨姆·奥尔特曼 (Sam Altman) 表示:“我认为我们正处于一个巨大模型时代的终结。我们会通过其他方式让他们变得更好。”
作者认为:
I don’t doubt that Inductive AIs can improve themselves. Given their own code, and their ability to write code, they could rewrite their code to be better. Induction, like deduction, can create new knowledge. But this kind of improvement is incremental, it can’t jump curves. Inductive intelligence can’t be used to conceptualize what I’ve called “Invention”, just as deductive intelligence could not conceptualize induction. This hard barrier to improvement is an impassable roadblock to the singularity. If you believe that human intelligence is something fundamentally more than induction, then we are not currently on the road to AGI.
Until we use our inventive intelligence to invent artificial Invention, the next curve jump, this AI can’t become AGI. And this assumes we even can: it may even be that humans can’t conceptualize our own intelligence. We can’t deduce deduction or induce induction, so why should we believe we can invent Invention? If we can’t, then the creation of the next kind of intelligence will probably only happen in the same way that our intelligence happened: not designed but chanced upon by trial and error through the process of evolution.
我毫不怀疑归纳式人工智能可以自我改进。考虑到他们自己的代码以及编写代码的能力,他们可以重写代码以使其更好。归纳法和演绎法一样,可以创造新的知识。但这种改进是渐进式的,不能跳跃曲线。归纳智能不能用来概念化我所说的“发明”,就像演绎智能不能概念化归纳一样。这种改进的硬障碍是通往奇点的不可逾越的障碍。如果你认为人类智能根本上不只是归纳,那么我们目前还没有走上通用人工智能的道路。
除非我们使用我们的发明智能(inventive intelligence)来发明人工发明(artificial Invention),才能完成下一个曲线跳跃,而现在的人工智能无法成为通用人工智能。这甚至假设我们有能力做到这一点:人类甚至可能无法概念化我们自己的智能。我们无法演绎演绎或归纳归纳,那么为什么我们应该相信我们可以发明发明呢?如果我们不能,那么下一代智能的创造可能只会以与我们的智能发生相同的方式发生:不是设计出来的,而是在进化过程中通过反复试验偶然发现的。

AI 也需要一个 inner coach。