随便写写。
progesterone hit me hard。上周，一点动力都没有。
我小时候经常觉得脑子里有很多个脑子。在思考一个东西的时候，总是一个脑子在想，但另一个脑子在想另一个，另一个脑子在想另一个，脑子一层叠着一层，想法一层叠着一层。长大后觉得脑子渐渐变成了一个，这一层脑子下面后的一层脑子已经听不清了。也许和大脑白质/灰质的变化相关。
小时候很少回忆过往的人生，可能是每天新奇的事物已经让人应接不暇，大脑还有足够的容量和成长的capacity。二十岁之后开始频繁回忆人生，拆检记忆，拾起又放下拼回。这让我很害怕，毕竟认知神经科学有说每当你把一片记忆取出再放去的时候其实都是在编辑记忆和给记忆上色，人的记忆很容易出错。每次我回想起一个记忆片段，我就担心我下一次记起来的时候这份记忆会不会离真实越来越遥远。我还会担心，会不会我这一次想起这个记忆片段后忘记放回去，我下一次就记不起来了。又或者，为什么在那一年的童年中发生了那么多事我却只想起了这样一件事，是不是其他我没有想起的事情已经要从我的脑子里消失了。我的记忆在膨胀压缩坍塌，我觉得我每天都在丢失一点记忆，丢失一部分的自己，当然，也可以是每天都在更新一部分的自己，构建和获得一部分的自己，只是遗忘这件事情，让我感到害怕，我怕我有一天会忘记我不想忘记的事情。可能是人的lost aversion的本能吧。
我的记忆每天都在做功，以抵抗时间的箭头和熵的增加。
小时候想发明一种传送带，从这个世界的任何一个角落，街道上任何一个地方，站到传送带上就能传到别的地方。
前几天在图书馆看到有些同学的课上reading，讲的是伊斯兰文化。上面说islam的意思就是submit，我的理解是将人submit给一个人无法主宰的地方。submit给一种很宏大的东西，感受自己的渺小，借此放下自己的痛苦或者是责任。我在游泳的时候福至心灵，我觉得游泳就是将我submit给水，我和我的水都很孤独，我的脑子一片空白，此时此刻拥有了这片水。
体育就是人的身体达到极限的时候去体验生命的一个过程。
喜欢跑步，当然也有endorphin kick in的时候整个人的身体和大脑会得到一种释放的感觉的因素。比这更喜欢的，是在身体达到极限的时候，每一次呼吸都变得非常得清晰，一秒钟的感觉变得像十秒，大脑似乎在有意识地控制和感受每个脚步和呼吸，我从未如此清晰地感觉到自己活着，驱使着我的双腿，身体承受着极大的痛，而在这种pain和suffering中生命的轮廓变得鲜活和清楚。生命的力量总在“濒死”的边缘感受到。我从未如此感谢自己的双腿，在最后的一公里里，我要像明天就要失去这双腿一样跑步。
I think Clawdbot is hype, as it is essentially a wrapper plus API access plus an LLM, and the real power here is still the LLM haha. for the consciousness part, we simply do not know and cannot tell. By the definition provided by the Turing test, I would say they are intelligent. whether they are conscious depends on how consciousness is defined.
When we refer to consciousness, we usually associate consciousness with subjective experience and awareness of the self, the body, and the environment. from a carbon-based mammalian perspective, all LLMs are silicon-based simulations of the computation of intelligence involving language, which is associated with parts of cognition, subjective experience, and awareness, and are designed to be aligned with the subspace of the communication channel that humans deploy. In this sense, it is a simulation of consciousness.
regardless, intelligence does not necessarily require the existence of consciousness. who can truly tell the difference between a simulation of consciousness and consciousness itself though. i myself am conscious and a simulation of consciousness at the same time.
I like the idea you mentioned the other day about new technologies liberating us by perhaps sacrificing some of our brain’s functionalities but allowing us to focus on what may be more important. LLM make some capability less significant, for example, basic coding, image editing, or writing LLM style garbage. At the same time, it makes other abilities more important, such as insight, aesthetic judgment, structural thinking, and decision-making. As someone who works closely with AI, I have never felt that it can replace critical thinking. Most of the comments it generates that feel “novel” or “critical” are ultimately derived from books written by people. However, what I truly like about is if someone is genuinely curious, LLMs can dramatically break down barriers of language and access to information. As long as you want to learn, you can keep asking questions, keep learning, keep discussing, and enjoy something close to free education, which is a huge privilege that was unimaginable for scholars in ancient time.
I believe everyone has an intrinsic curiosity about the world when they are children, and our education system should encourage this kind of self-driven motivation. Too much external reward can undermine internal motivation. Kant once argued that when people believe in God, it becomes difficult to tell whether their goodness is genuine or merely instrumental(done for the reward of going to heaven), thus undermining free will. I think the same applies to the pursuit of knowledge. If someone has a pure desire to learn, LLMs can satisfy and support that curiosity. But if everything is driven by instrumental goals within an evaluation system, then LLMs may actually hinder real learning. But that’s not the fault of LLM, rather, the mindset of ppl who use them this way; or rather, the system that nudge ppl to use them this way.
