Behind the Hype of Sora, What's Truly Worth Learning?
-
It's been over half a month since Sora was released, and the industry's discussions show no signs of cooling down.
But as we delve deeper into this 'product' (if it can even be called that yet) and understand its underlying principles, a counterintuitive question arises—is Sora really worth the significant time investment for ordinary people to 'learn'?
Currently, the so-called 'learning materials' surrounding Sora mainly focus on a few dimensions: First is the technical aspect, exploring the underlying implementation principles. Whether it's tech gurus, PhDs in academia, or tech media influencers, the discussions essentially revolve around: "Diffusion Transformer + Spatiotemporal Autoencoder + SpaceTime patches + Scaling Law". Speaking too superficially invites disdain from experts, while going too deep loses audience interest. The key question is: after understanding these principles, what's next?
Second is the philosophical aspect, examining whether we've achieved a world model, whether the real world might be overturned, and whether we're getting closer to the AGI era. But such open-ended discussions have no boundaries - different parties debate with their own arguments while spectators watch the show. Again, what's next? The third aspect is traffic chasing - teaching you how to make money with Sora, where to find Sora videos, how to leverage keywords for traffic, how to write Sora prompts, and how to apply for Sora beta access. There's also the practice of using Sora topics to create anxiety and sell courses or build communities. But after looking around, apart from those who were already skilled at making money ('operators'), I haven't heard of anyone truly benefiting from learning about Sora.
This article isn't discussing the right or wrong of these phenomena - I'm also working in AI and have ridden this trending topic. But beyond that, I want to discuss perspectives outside the Sora hype, namely:
Looking at it from the end goal, in the grand trend of AI explosion, what should ordinary people really learn? I think there are three key points: storytelling ability, structured thinking, and content expertise.
Taking text-to-video generation as an example, not to mention Sora, many people can't even use Pika or Runway proficiently.
Describing an event, an opinion, or a request in words is not difficult, but describing a scene, a setting, or a complete story is a high barrier for the vast majority of people. Many people worry that the film industry will be disrupted in the future, believing anyone could use Sora to create blockbusters. I don't believe this.
As a content product manager with years of experience, I firmly believe creating good works requires repeated learning and extensive training in writing, painting, and expression. To effectively use AI, one must first thoroughly study works like The Hero's Journey and Save the Cat, and analyze dozens of classic films just to get started. I recommend reading my previous article: [Reading 681. Learning How to Tell Stories - Reflections on Save the Cat]
Sora won't make the film industry obsolete. On the contrary, as AI video generation technology becomes more widespread, it will allow those who truly know how to craft stories to showcase their talents, rather than being discouraged and giving up due to the industry's unwritten rules. Many articles about Sora focus on its innovations in model algorithms, computational power emergence, and physical simulation. For me, what's more astonishing is how it addresses the video annotation problem during training by employing a structured prompt method:
"We first train a highly descriptive captioner model and then use it to produce text captions for all videos in our training set"
At its core, structured expression capability is utilized in both B-end training data and C-end interactions with large models. Even those so-called "Sora training" courses essentially teach you how to write effective prompts. For example, consider this scenario:
How should it be described? A thoughtful old man? An old man wearing glasses? An old man daydreaming in a café? Most people might stop there. But a more structured description could be: Main Character: A white-haired, bearded man in his 60s
Shot Type: Close-up of the face
Background Setting: A café in Paris Character State: Gazing out the window at the passing people, motionless in deep thought, then smiling slightly with pursed lips.
Character Details: Wearing a wool coat and a button-up shirt, with a brown beret on the head and glasses, resembling a professor.
Environmental Details: Golden lighting with Parisian streets and the city in the background. Lens Technology: Depth of field, 35mm cinematic film for movie-like effects
The ability to craft such prompts is crucial for unlocking the value of multimodal technologies. This fundamentally requires structured summarization and analytical thinking skills. Daily Reflection 396: How to improve summarization abilities?
There's an apt saying: Writing prompts is like clients giving a brief - being able to clearly articulate requirements means you're already halfway to the correct answer. This brings us back to the creators themselves. I particularly agree with the perspective shared by Yang Yuancheng (co-founder of "Tang Island" and "New World Phase") in his article After the Sora Hype, Rationality Emerges | I Asked 5 Top Creators for Their Real Thoughts:
The biggest impact will be the emergence of new creators—content idols who were previously outside the public eye. They possess excellent taste,思维方式, and worldviews but were limited by technical barriers, lacking even basic video production skills in the past.
Yes, beyond the storytelling ability mentioned earlier. If you have a deep understanding of what constitutes "good content" and know how to design scripts, storyboard, apply effects, frame shots, direct scenes, cast actors, set lighting, dub, and edit post-production—AI becomes your best ally in unleashing creativity. Those who can effectively utilize AI will undoubtedly create better works and achieve greater financial success. Therefore, I firmly believe that this wave of AI trends is an absolute boon for high-potential creators such as writers, artists, designers, and screenwriters. If you also want to benefit from it, start by setting "creating a great piece of content" as your learning goal.
Overall, I think there's little need to chase news about Sora. Instead of getting caught up in the overwhelming hype around Sora itself, focus on mastering the essential skills behind it. Train your storytelling ability with structured thinking and become an outstanding creator—that’s what’s truly worth investing your time in.
I’m Shen Yue, a former product expert at a major tech company and now an entrepreneur and observer in the AI field. I look forward to exchanging ideas with you.