读前提醒,本文只代表作者个人观点。
========
有一说一,我焦虑了。不是因为模型太强可能把自己替代什么的,而是从频繁的模型服务器过热和不断降低的额度中意识到了,可能来自新时代的第一轮的筛选已经到来了。
如果说未来人的认知会被套上ai这层外骨骼,那么一个可以使用sota模型的人和不能的人之间的差距恐怕会仅仅因一个模型而远远拉开。
而我焦虑的就在这里,如果说为了使用sota模型需要花上100u,恐怕大部分人还是可以做到/或者说愿意花的;但是如果说当前的价格只是在补贴后的价格,那么当这个价格来到200u 400u的时候,为了一个不是一定可以提升自己生活的模型,大部分人还愿意花上这么高昂的价格作为自己生活的必须吗。那么这里就是第一道的筛选,用单纯的金额,就可以把人分成可以使用sota模型的人和不行的部分。
虽然感觉在现在应该算是ai热潮的主升浪的阶段说这个话还有点为时尚早,不过现在这个时间点,也可能说算是右侧的习惯吧(
至于说,为什么自己会想到这个,大概是因为mythos。自己看到它在安全领域部分的作用,然后这大概刺到了我,让我意识到他们其实并不是只满足在coding这一块领域。
然后是这份刺激让我回想到了他们在视频领域的工作和推广,在作曲领域的尝试。其实他们早就不止满足于代码领域了,那么当一个产业从推广进入了平台期,对于用户而言的使用成本还会像现在这么低廉吗。
作为没有什么本金的人,只能被时代推着,向前摸索,不知对错,或许确实悲哀;不过,面临新世界的机会,同时也是平摊在所有人面前的,或许说这也是一种来自时代的馈赠。
Before you read, a reminder: this article only represents the author’s personal views.
========
To be honest, I’m anxious. Not because the models are so strong that they might replace me or anything, but because from the frequent overheating of model servers and the continuously shrinking quotas, I’ve realized that the first round of selection from this new era may already have arrived.
If, in the future, human cognition will be fitted with an AI exoskeleton, then the gap between someone who can use SOTA models and someone who can’t will likely be widened enormously—just because of a single model.
And that’s where my anxiety lies. If using a SOTA model costs 100u, most people can probably manage that / or would be willing to pay it. But if today’s price is only the subsidized price, then when it becomes 200u or 400u, will most people still be willing to pay such an exorbitant price for a model that isn’t guaranteed to improve their lives, treating it as a necessity? This is the first filter: with money alone, people can be split into those who can use SOTA models and those who can’t.
Although it may be a bit early to say this at a time that probably still counts as the main upward wave of the AI boom, at this point in time it might also be considered a “right-side” habit, I guess (
As for why I thought of this, it’s probably because of Mythos. I saw what it can do in parts of the security domain, and that probably hit a nerve, making me realize they aren’t actually satisfied with just the coding domain.
Then that jolt made me think back to their work and promotion in the video domain, and their attempts in composition. In fact, they stopped being satisfied with only the code domain a long time ago. So when an industry moves from promotion into a plateau phase, will the cost of use for users still be as low as it is now?
As someone without much capital, I can only be pushed forward by the times, groping ahead, not knowing right from wrong—perhaps it is indeed tragic. However, the opportunities of the new world are also spread out in front of everyone equally; perhaps this, too, is a kind of gift from the times.
读前提醒,本文只代表作者个人观点。
========
说实话,我焦虑了。不是因为模型太强、可能会把自己替代之类的,而是从频繁的模型服务器过热和不断降低的额度中意识到:或许来自新时代的第一轮筛选已经到来了。
如果说未来人的认知会被套上一层名为 AI 的外骨骼,那么能使用 SOTA 模型的人和不能使用的人之间的差距,恐怕仅仅因为一个模型就会被远远拉开。
而我焦虑的点就在这里:如果说为了使用 SOTA 模型需要花 100u,恐怕大部分人还是做得到/或者说愿意花的;但如果说现在的价格只是补贴后的价格,那么当这个价格来到 200u、400u 的时候,为了一个未必一定能提升自己生活的模型,大部分人还愿意花这么高昂的价格,把它当作生活必需品吗?那么这里就是第一道筛选:只凭金额,就能把人分成能使用 SOTA 模型的人,以及不能的那部分。
虽然感觉在现在这个应该算是 AI 热潮主升浪的阶段说这话还有点为时尚早,不过在这个时间点,也可能算是右侧的习惯吧(
至于说,为什么自己会想到这个,大概是因为 mythos。自己看到了它在安全领域一部分的作用,然后这大概刺到了我,让我意识到他们其实并不只满足于 coding 这一块领域。
然后这份刺激让我回想起他们在视频领域的工作和推广、在作曲领域的尝试。其实他们早就不止满足于代码领域了;那么当一个产业从推广进入平台期,对于用户而言的使用成本还会像现在这么低廉吗?
作为没有什么本金的人,只能被时代推着向前摸索,不知对错,或许确实悲哀;不过,面临新世界的机会同时也是平摊在所有人面前的,或许这也是一种来自时代的馈赠。
部分信息可能已经过时









