科学家让人工智能狂看抖音小姐姐跳舞来学习？网友：这应该就是以后机器人起义的原因吧
How TikTok dances trained an AI to see
译文简介
Vox发布Phil的人工智能系列视频之四：AI如果通过观看抖音舞蹈来学习
关于这项研究，有些网友看好，表示用这种方法训练AI非常明智，是科学与艺术的碰撞；也有部分网友担忧：万一机器人起义怎么办？抖音不是中国收集我们数据的阴谋吗？我们的数据被用于研究还尚可，但如果公司资助这些研究并以此获利，我们的权益如何得到保障？
正文翻译
我们可以用抽认卡片来比喻人工智能（AI）的学习过程：卡片的正面是问题，反面是答案，电脑可以以此获得正确的内容。要想让AI学习3D人体的静态与动态，静止挑战的图片库和抖音跳舞视频就是答案。从最初建立3D模型网站，到通过程序将抖音舞蹈视频抽象成3D视频并进行观测，再到以大量静止挑战视频作为数据集，“抽认卡”的数量越来越庞大。
注：静止挑战，直译为人体模型挑战，人们假装人体模型摆出各种造型，呈静止状态，然后用手持相机移动拍下整个现场。
评论翻译
很赞 ( 3 )
收藏
Forcing an AI to watch countless hours of TikTok is definitely how the robot uprising starts
逼AI狂看抖音一定会成为机器人起义的原因
原创翻译：龙腾网 http://www.ltaaa.cn 转载请注明出处
The AI must be traumatized
AI肯定会精神崩溃
It will help with that!
Now we're closer to getting 99% accurately shot at by robot soldiers! ^u^
对这个肯定有帮助：
现在我们离被机器人士兵99%精确射击又更进一步啦！ ^u^
And now we cannot even blame them.
而且现在我们甚至都没法怪它们。
Using old Mannequin Challenge videos for training a depth prediction model is such a wildly clever idea!
用静止挑战的老视频来训练深度预测模型这个主意真是太聪明辣！
i really like how phil is so personable, it feels like he's casually having a conversation with just me. a great presenter
Phil（视频博主）太有风度了，我好爱。感觉就像他在和我单独唠家常一样。很棒的博主
That’s actually insanely cool. A bizarre social trend from years ago becomes the weirdly perfect dataset for these computer vision models. Those researchers literally couldn’t have asked for better
实际上真的是酷毙了。多年前这个奇奇怪怪的媒体潮流（指静止挑战）居然莫名其妙地成为了这些计算机视觉模型的完美数据库。研究者们确实找不到比这更适合的了
Plot twist, they were the ones who created the trend in the first place
Plot twist，他们是最早一批制造这个潮流的
@cheesybirdmess7349 This is what I was thinking about. But in reverse: What trend should we start next to create another, even more amazing dataset?
@cheesybirdmess7349 我也是这么想的。但反过来说：我们下一个该制造些什么潮流来创造出另一个甚至更棒的数据集呢？
I believe that the challenges are a way to collect data to train AI. Not just TikTok, People share photos from their childhood, teens and adult versions. No wonder how good those filters work
Some challenges are just in line with AI data projects that I’ve rolled in.
我觉得静止挑战是个收集数据训练AI的不错办法。人们可不止在抖音上分享（视频），他们还分享自己的童年，青少年和成年照片。难怪那些滤镜效果那么好
有些挑战者（的数据）刚好和我投资的AI数据项目吻合。
I'm loving these AI-themed videos from Phil!
我爱Phil做的这些AI主题的视频！
原创翻译：龙腾网 http://www.ltaaa.cn 转载请注明出处
ikr, Its fascinating but also scary at times.
赞同，视频很吸引人，但有时也很吓人。
It's scary to think that AI companies can use data like this even if it's to train an AI. I guess it's the same way with how image generators like Bluewillow was trained with a few differences.
想想AI公司能使用这样的数据，即使是为了训练AI也很吓人。我猜它的训练方式跟Bluewillow这样的图像生成器差不太多。
Though the methodology is fascinating, the actual question is what will this research be applied on? I can imagine predictive vigilance and ever-expanding tracking and warfare capabilities. It'd also be nice to know who has funded all of these researches.
尽管这项方法论很吸引人，但实际的问题是这项研究将应用于何处？我能想象到对此的预测性警戒和不断扩张的跟踪以及战争可能。要是能知道谁在资助所有这些研究也挺不错。
The whole point of tiktok has ways been AI. People fear China getting stronger than us, but you’re helping them every time you use that app. Great job.
抖音的重点从来都是AI。人们害怕中国变得比美国强，但是每次你用这个软件的时候，你都在帮他们。干得漂亮。
Never thought I'd see a day when the mannequin challenge becomes useful.
从未想过有一天我能看到静止挑战这么有用。
Imagine if your entire purpose in life was to watch tiktok dance videos endlessly. When the robot uprising happens I bet this one will be extra cruel in return for what we made it do
想象一下你生活的全部目的就是无休止地看抖音的跳舞视频。当机器人起义的时候，作为我们让它们看视频回报，这个惩罚肯定格外残忍。
I never downloaded tiktok way back when it was being advertised a ton on Snapchat because my assumption was that it was the “dance” app was going to be used to aid AI in understanding body movement for the Chinese government to further improve their I.D. tools on an individual’s gait and other identifiers. This just further reinforces that the original purpose was not a fun dance app.
当抖音在Snapchat上铺天盖地打广告的时候我一直没有下载，因为我觉得这个“跳舞”软件会被用于辅助AI来了解人体的身体运动，这样c就能通过某个人的步态和其他特点来进一步完善他们的身份验证工具。这只会进一步说明它本来就不是一个跳舞的娱乐软件
Everything is a training set. You use to be an advertising target, now you are a training set.
一切都可以作为（AI的）训练数据集。以前可以打广告，现在可以当训练数据集。
As people have pointed out in the comments: we don't have to accept that our personal data, our images and videos can be used by anyone for any purpose. Doing research is great In many countries it is allowed to use most data for that purpose. The problem begins when said research is sponsored by a for profit company and is used by them for generating profit. That's a loophole and it's been abused for the last years. AI image generators like Stable Diffusion, Midjourney and the others have abused this loophole to create their product that's been developed on the data they have no rights to. No matter how cool the technology is, we have to protect our rights!
就像有些评论里指出的：我们不必接受个人数据，个人形象和视频被任何人以任何目的来使用。在很多国家做研究是很好的，他们允许以研究为目的使用大部分的数据。如果研究是为盈利性公司所资助，并且成果被公司用来盈利的时候，问题就会开始显现。这是个漏洞，而且已经被滥用了好几年。像Stable Diffusion, Midjourney等等这样的AI图像生成工具已经利用这个漏洞来创造基于他们有版权的数据的产品了。无论这项技术有多好，我们必须保护自己的权利！
Imagine if this video was AI generated and it turned out Vox was never real to begin with
想象一下这个视频是AI生成的，而且Vox（up主）一开始就不存在
It'd be nice if we could all agree that the term "AI" will be applicable once they can teach themselves. In the meantime, "ML" for "machine learning" is the same number of letters.
一旦人工智能开始自学，“AI”这一术语就变得名副其实，要是我们都同意这一点就好了。与此同时，“机器学习”变成“ML”，也是两个字母
I don't even like tiktok or dances, challenges but this is wild. Also, great video guys
我甚至都不喜欢抖音和跳舞，也不喜欢静止挑战，但这真的很厉害。铁子们视频很棒哟
我在一项艺术创作中使用了AI，它被训练用于检测场景的3D深度估计。
猜猜它是通过什么来训练的：
数据集很多，但主要是：
汽车监控摄像拍下的街景
我们有足够的数据来让AI猜测下一帧画面是怎样“向前推进”的（物体是接近还是远离），然后检测它的精确度。
Don’t most people post things online to communicate with friends or have fun? When did we sign up for collecting data and training machines?
大部分人在网上发布内容不是为了和朋友交流或者好玩吗？我们什么时候报名参加数据收集和机器训练了？
It's really cool to see my current research direction summarized in a video. :D
能在视频里看到别人总结我现在的研究方向真是泰裤辣 :D
Is beautiful when art meets science
艺术与科学的碰撞是如此美妙