If you said the one on the right, congratulations, you're correct.

如果你說的是右邊那個，那麼恭喜你，你說對了。

But you would also be correct if you said the one on the left, because both of these videos were made by AI.

但如果你說左邊那個也是對的，因為這兩個視頻都是人工智能製作的。

And ever since VO2 was released on December 16th, I've been asking myself, am I replaceable?

自從 12 月 16 日《VO2》發佈以來，我一直在問自己：我是可以被取代的嗎？

Were the years I've spent learning Blender and Videography a massive waste of time?

我花了這麼多年學習 Blender 和攝像，難道是在浪費時間嗎？

And I'm not alone.

我並不孤單。

It's estimated that 300 million jobs could be replaced by AI in the next 5 years.

據估計，未來 5 年內將有 3 億個工作崗位被人工智能取代。

So as 3D artists, what AI should we be worried about?

那麼，作為 3D 藝術家，我們應該擔心哪些人工智能呢？

There are three main types of AI, each with unique dangers and benefits.

人工智能主要有三種類型，每種類型都有獨特的危險和好處。

I'm sure you're familiar with type 1.

我相信你對 1 型並不陌生。

These are video and image generation AI, the best of which can produce images that look like this, and videos that look surprisingly realistic.

這些都是視頻和影像生成人工智能，其中最好的人工智能可以生成看起來像這樣的影像，以及看起來驚人逼真的視頻。

For video, Google's VO2 is the best right now, with Sora as a distant second.

在視頻方面，Google 的 VO2 是目前最好的，Sora 遙遙領先。

Image generation is a closer race, with Dolly 3 and Mid Journey both able to produce convincing results.

影像生成方面的競爭更為激烈，Dolly 3 和 Mid Journey 都取得了令人信服的成績。

In their current form, these prompt-based AI have already replaced many artists, specifically by taking away smaller clients.

目前，這些基於提示的人工智能已經取代了許多藝術家，特別是搶走了小客戶。

I mean, you're not going to see VO2 in the next Dune movie.

我的意思是，你不會在下一部《沙丘》電影中看到VO2。

They may use AI tools, but they're not going to use fully AI generated imagery.

他們可能會使用人工智能工具，但不會使用完全由人工智能生成的影像。

I'm sure they would if they could get away with it, but AI shots aren't consistently realistic, and so many people would boycott the movie that I don't think we'll see fully AI generated imagery in blockbusters anytime soon.

如果可以的話，我相信他們會這麼做的，但人工智能的鏡頭並不總是那麼逼真，而且很多人會抵制這部電影，所以我不認為我們很快就能在大片中看到完全由人工智能生成的影像。

But for many smaller companies, that's not the case.

但對於許多小公司來說，情況並非如此。

Just put yourself in the shoes of a business owner.

設身處地為企業主著想吧。

If you're trying to get some new content for your new website or ad campaign, are you going to choose the option that is virtually free, has a turnaround time of minutes, and gets you 80% realism?

如果你想為你的新網站或廣告活動獲取一些新內容，你會選擇幾乎免費、幾分鐘就能完成、並能獲得 80% 真實度的選項嗎？

Or spend thousands working with videographers, 3D artists, and web designers for weeks on end to get to that 100% mark?

或者花費數千元與攝像師、3D 藝術家和網頁設計師合作，連續數週達到 100% 的目標？

It's a tough decision, especially if those extra savings could go into a larger advertising budget or grow your business.

這是一個艱難的決定，尤其是如果這些額外節省的資金可以用於更多的廣告預算或發展業務。

I mean, I can see the dilemma.

我的意思是，我明白你的兩難處境。

So AI generated imagery is dangerous right now, and there's no telling where it will go in the future.

是以，人工智能生成的影像現在很危險，未來會走向何方還不得而知。

But type 2 AI is both more dangerous and more useful for artists.

但對於藝術家來說，第二類人工智能既更危險，也更有用。

These are AI tools.

這些都是人工智能工具。

They don't generate entire videos themselves, instead they help with certain parts of production.

他們不會自己製作整個視頻，而是幫助製作某些部分。

This can be extremely beneficial, but it's also dangerous.

這可能非常有益，但也很危險。

So what are some of the most powerful AI tools?

那麼，有哪些最強大的人工智能工具呢？

Well first is Blend AI.

首先是混合人工智能。

This is a prime example of how AI can be helpful for creatives.

這就是人工智能如何幫助創意人員的最好例證。

Blend AI is an add-on that essentially gives you an AI personal assistant for how to execute technical tasks inside of Blender.

Blend AI 是一個插件，本質上是為你提供一個人工智能私人助理，幫助你在 Blender 中執行技術任務。

It has some other features as well, and this is an all-around awesome application of AI.

它還具有其他一些功能，是人工智能的全面出色應用。

But next is something far more dangerous. 3dfy.ai is a platform that allows you to create 3D models based on a prompt.

但接下來的事情要危險得多。3dfy.ai 是一個可以根據提示創建 3D 模型的平臺。

Now right now, there is nothing to worry about.

現在，沒什麼好擔心的。

This was a feature that was implemented into many Autodesk programs about a year ago now, and no one uses it.

大約一年前，許多 Autodesk 程序都採用了這一功能，但卻無人使用。

The models it creates are sloppy, and they need a ton of work to redepologize.

它創建的模型馬馬虎虎，需要做大量的工作來重新道歉。

So why is it on this list?

那麼，它為什麼會出現在這份名單上呢？

This demonstrates the fundamental philosophy that the industry cares about three things.

這體現了業界關心三件事的基本理念。

Cares about results, time, and money.

注重結果、時間和金錢。

AI is already fast and cheap, so once it gets to the point where it can deliver good results as well by creating models of good topology, that's going to take away a lot of jobs.

人工智能的速度已經很快，成本也很低，所以一旦它能夠通過創建拓撲結構良好的模型來提供良好的結果，就會搶走很多工作。

This trend is continued with the next AI, Blender GPT.

下一個人工智能 Blender GPT 將延續這一趨勢。

This is another add-on for Blender, and it has some use cases.

這是 Blender 的另一個插件，也有一些用例。

But unlike Blend AI, it's far more dangerous than it is useful.

但與 Blend AI 不同的是，它的危險性遠遠大於實用性。

Essentially Blender GPT allows you to type a prompt, and it will take that prompt and write a Python script that will then add whatever you ask it to create as 3D objects in your scene.

從本質上講，Blender GPT 允許你輸入一個提示，它會根據提示編寫一個 Python 腳本，然後將你要求它創建的任何內容添加到場景中的 3D 對象中。

So why is Type 2 AI so much more dangerous than Type 1?

那麼，為什麼 2 型人工智能比 1 型人工智能危險得多呢？

So first, why is AI dangerous in the first place at all?

那麼，首先，為什麼人工智能會有危險呢？

Well, specifically for creatives, it's dangerous because it can replace you, take your job.

特別是對創意人員來說，這很危險，因為它可以取代你，搶走你的工作。

That's only going to happen if it's widely adopted.

只有廣泛採用，才能實現這一點。

Now we talked about why many large studios won't use fully generated AI images, because of the results, but also because of public opinion.

現在，我們談到了為什麼許多大型工作室不會使用完全生成的人工智能影像，這不僅是因為效果，還因為公眾輿論。

AI tools don't have the same stigma as fully generated AI shots.

人工智能工具並不像完全生成的人工智能鏡頭那樣具有恥辱感。

And they also yield better results.

它們還能產生更好的效果。

Because they aren't replacing the entire studio, instead they're replacing the modeling team, or the animation department.

因為他們更換的不是整個工作室，而是建模團隊或動畫部門。

And as AI continues to get better, we may see artistic integrity disappear.

隨著人工智能的不斷進步，我們可能會看到藝術完整性的消失。

As artists are forced to either use AI to meet insane deadlines, or they're laid off.

藝術家們要麼被迫使用人工智能來趕工期，要麼被解僱。

And that's a scary future.

那是一個可怕的未來。

And there's one AI that hasn't been released yet that could bring that future into reality.

有一種尚未發佈的人工智能可以將未來變為現實。

And that's Meta's XR project.

這就是 Meta 的 XR 項目。

This is an attempt to create fully immersive virtual and augmented reality worlds that are entirely powered by AI.

這是一種嘗試，旨在創建完全由人工智能驅動的沉浸式虛擬和增強現實世界。

They're going to be doing this to implement this technology into their VR headsets in the Metaverse, but also into their new AR glasses as that technology continues to develop.

他們將把這項技術應用到 Metaverse 的 VR 頭顯中，同時也會應用到新的 AR 眼鏡中，因為這項技術還在不斷髮展。

However, the problems that need to be solved to create immersive virtual worlds are the same problems that are preventing AI from taking over the film industry.

然而，創造身臨其境的虛擬世界需要解決的問題，也正是阻礙人工智能接管電影業的問題。

Things like realistic model and texture generation, powerful AI-driven animation, and increased render performance to be able to work in real time.

如逼真的模型和紋理生成、強大的人工智能驅動動畫，以及更高的渲染性能，以便能夠實時工作。

Meta is planning on launching this in early 2026, but that doesn't mean that there are going to be huge layoffs right away.

Meta 公司計劃在 2026 年初推出這款產品，但這並不意味著馬上就會大量裁員。

But slowly, AI will take over many of the tasks that were once performed by artists.

但慢慢地，人工智能將取代許多曾經由藝術家完成的任務。

And that's where Type 3 AI comes in.

這就是第三類人工智能的用武之地。

Type 3 is the most dangerous AI.

第三類是最危險的人工智能。

But it's not here yet.

但現在還沒到。

OpenAI announced their plan for AI agents, and its intended release date is this month, January 2025.

OpenAI 宣佈了他們的人工智能代理計劃，計劃的發佈日期是本月，即 2025 年 1 月。

So what is an AI agent?

那麼，什麼是人工智能代理呢？

How could it be helpful?

怎麼會有幫助呢？

And why is it so dangerous?

它為什麼如此危險？

The whole idea of an AI agent, also known as an operator, is that you have your own personal AI that's living on your computer and can do basically everything you can.

人工智能代理（也稱操作員）的整體概念是，你擁有自己的個人人工智能，它就住在你的電腦上，基本上能做你能做的一切。

It can work with text, code, images, videos, and you essentially just have to say, book me a dinner reservation, send out an email to my co-workers, or start a project file in Blender.

它可以處理文本、代碼、影像和視頻，你基本上只需要說，給我訂一個晚餐，給我的同事發一封郵件，或者在 Blender 中啟動一個項目文件。

Because agents live on your computer, they have access to your programs and can essentially do anything you can.

由於代理生活在你的電腦上，他們可以訪問你的程序，基本上可以做你能做的任何事情。

But that is the ultimate vision.

但這才是最終的願景。

Right now, we don't have AI agents, but there are two that are coming soon that are going to be complete game changers.

現在，我們還沒有人工智能代理，但有兩個即將推出的代理將徹底改變遊戲規則。

First is AI's multi-modal agent that is due to release this month.

首先是人工智能的多模式代理，將於本月發佈。

It won't be able to open other programs like Blender, but it's optimized for problem solving and artistic creation through its own system.

它無法打開 Blender 等其他程序，但它通過自己的系統優化了解決問題和藝術創作的能力。

But this is the first iteration, the first generation of AI agents.

但這只是第一次迭代，是第一代人工智能代理。

So in future releases, they're only going to get better.

是以，在未來的版本中，它們只會變得更好。

Next is ServiceNow's AI, which is focused on integration with other platforms and then utilizing the integration for workflow automation.

其次是 ServiceNow 的人工智能，它的重點是與其他平臺集成，然後利用集成實現工作流程自動化。

They're essentially trying to bridge the gap between consumers and corporations by replacing customer service teams and sales staff with AI agents.

從本質上講，他們試圖用人工智能代理取代客戶服務團隊和銷售人員，從而在消費者和企業之間架起一座橋樑。

And this integration tech is becoming more and more powerful.

而這種集成技術正變得越來越強大。

Just think about these agents.

想想這些特工吧。

They work for you 7 days a week, 24 hours a day, 365 days a year, and you don't have to pay them or take care of them.

他們一週 7 天、一天 24 小時、一年 365 天為你工作，你無需向他們支付報酬或照顧他們。

And in our case, they can run your IT department, your customer experience, your employee experience.

在我們的案例中，他們可以管理 IT 部門、客戶體驗和員工體驗。

ServiceNow has partnered with Nvidia, IBM, and Microsoft to create an even more powerful integration system.

ServiceNow 與 Nvidia、IBM 和 Microsoft 合作，創建了一個功能更加強大的集成系統。

Bill McDermott described ServiceNow as the control tower that integrates other AI systems.

Bill McDermott 將 ServiceNow 描述為整合其他人工智能系統的控制塔。

This means as the integration network and agent technology continues to develop, this will become even more revolutionary, but also more dangerous.

這意味著，隨著集成網絡和代理技術的不斷髮展，這將變得更加革命性，但也更加危險。

So what jobs specifically are in danger?

那麼，具體是哪些工作面臨危險呢？

And how do you protect yourself?

你如何保護自己？

Well, there are three visual effects positions that are most at risk.

有三個視覺特效崗位的風險最大。

We'll start with the least and go to the most in danger.

我們將從最不危險的開始，到最危險的為止。

First are texture artists.

首先是紋理藝術家。

Texture work requires a lot of tedious technical knowledge, like UV editing, that is on track to be replaced by AI.

紋理工作需要大量乏味的技術知識，如 UV 編輯，而這些都將被人工智能所取代。

Now texturing is also largely creative, with many artistic choices leading to the final result, which is why it's not going to be entirely replaced.

現在，紋理製作在很大程度上也是一種創造性的工作，許多藝術性的選擇會導致最終的結果，這也是為什麼紋理製作不會被完全取代的原因。

Second are compositors.

其次是合成師。

We've already seen After Effects' Rotobrush, that utilizes AI to rotoscope more effectively, and Adobe's FastFill project, which is better at paintouts than many artists.

我們已經見識過 After Effects 的 Rotobrush，它利用人工智能更有效地實現了旋轉效果，還有 Adobe 的 FastFill 項目，它在繪畫方面比許多藝術家都要出色。

But as these tools continue to improve, demand for talented compositors will decrease significantly.

但隨著這些工具的不斷改進，對優秀合成師的需求將大幅減少。

Now I'm not trying to sensationalize this, there's still going to be a need for compositors.

我並不是想聳人聽聞，我們仍然需要合成師。

But even if demand drops 20%, that's still a huge blow.

但是，即使需求下降 20%，這仍然是一個巨大的打擊。

Third is concept artists.

第三是概念藝術家。

This is a field that has already been encroached on by Dolly 3, VO2, and other AI models.

多莉 3 號、VO2 和其他人工智能模型已經蠶食了這一領域。

But as they continue to improve, the need for a human concept artist, or previs tech, drops.

但隨著它們的不斷改進，對人類概念藝術家或預視技術的需求也在下降。

Because in previs, you don't need really high quality or realistic results, you're just trying to block out the main story and animation.

因為在預視中，你並不需要真正高質量或逼真的效果，你只是想遮擋住主要故事和動畫。

Finally, Blender itself is planning on adding AI tools.

最後，Blender 本身也計劃添加人工智能工具。

So if you care about that, you can watch this video here.

是以，如果你關心這個問題，可以在這裡觀看這段視頻。