What Miyazaki Got Wrong About AI
The legendary animator dismissed AI as soulless. But he missed something crucial about how creativity actually works.
Man vs. Machine: Defining AI's Role in Content Creation
Back in 2017, a team of animators at Studio Ghibli excitedly presented an AI-generated animation to their boss - legendary filmmaker Hayao Miyazaki.
He was less than impressed.
"I am utterly disgusted," Miyazaki said, his face stern. "If you want to make something creepy, go ahead. I never wish to incorporate this technology into my work. I strongly feel that this is an insult to life itself."
The young animators looked crestfallen.
"We humans are losing faith in ourselves," Miyazaki added.
When grilled by another senior executive about their goal, one young animator piped up:
"Well, we would like to build a machine that can draw pictures like humans do."
Miyazaki's response was grim. "I feel like we are nearing the end of times."
Today, we are 7 years closer to the end of times. OpenAI's "Sora" video generation model not only can draw pictures like humans do, but much more.
Meanwhile, the head of NVIDIA says that humans need not learn to program anymore, since computers will increasingly do it for us.
What Miyazaki Gets Right
AI is no replacement for human creativity.
However, we can still recognize it as a powerful lever for aiding and enhancing human creativity and efficiency.
Especially when it comes to functional "jobs to be done," AI can act as a powerful tool, amplifying human creativity and efficiency. This is especially clear in the realm of programming and computers, where software amplifies our cognitive faculties.
While certain programmers look at their work as art, few are crying over the prospect of programming jobs going away. It's the fear of human creativity and values being stripped out of the process by cheap, mass-produced AI-driven content.
The Question of Content
Miyazaki's revulsion to AI in artistic processes raises an important question: What is the role of AI in creating what now goes by the ubiquitous name of Content?
People talk about digital content like a homogenous blob. Pre-internet, it was clearer that content served specific purposes. Films, newsletters, books, music, and even how-to manuals, SOPs, and other functional documents have always been created to serve a particular function: to do a job.
The end goal and the job to be done matter.
There are many purposes for content - from emotional, to financial, to functional. The common denominator is the satisfaction of a human need or desire. This has become less clear in 2024, when certain forms of content seem to exist for their own sake. People are looking to monetize eyeballs and attention without regard for underlying value.
One Law for AI Content
We can't argue with Miyazaki about what makes an artistic masterpiece, but we can lay out definitive criteria for the value of AI in content creation.
Asimov had his three laws for robots; we just have this one:
If an AI-generated work resonates with its intended audience and fulfills their needs, it has accomplished its purpose.
It's similar to how we perceive the value of currency. If a group of people agree that something has value, whether abstract or tangible, then it is deemed valuable.
To some, this criteria may sound too value neutral. By this definition, wouldn't the endless streams of AI-driven clickbait count as valuable?
But the answer is no. They might provide value to the creator in the form of ad revenue, but if it's not helping to get a job done, it violates our criteria.
The Fallacy of the Great AI Replacement
The fallacy behind common fears of the Great AI Replacement is thinking that AI is like a magic wand that can wave away the need for human input.
"a machine that can draw pictures like humans do."
The truth is, it's a tool that requires careful guidance and context to produce meaningful results.
Sora's 8-second video clips are impressive. But the real value will come from the way animators and storytellers string together AI prompts in ways that captivate our attention and deliver inner meaning.
In the realm of AI-assisted writing, the creator must seed AI with relevant context and source material. The quality of the output depends heavily on the quality of the input.
Garbage in, garbage out.
Treasure in, treasure out.
Transcripts at the Heart of the Workflow
That's why I put transcripts at the heart of my workflow. Conversations lay out the context and ideas that inform AI's outputs, ensuring results are relevant, useful, and aligned with audience goals. By providing clear instructions, carefully crafted prompts, and regular feedback, we can steer AI towards producing better, more relevant results.
When it comes to functional writing and content creation - Jobs to Be Done - AI can act as a powerful lever. I've experienced this firsthand, turning podcasts into show notes, summaries, and even articles, newsletters, and books written in the voice of the host. What used to take three to four hours for a podcast, I can now do in 30 minutes, with maybe 15-20 more minutes of polishing.
The Key Insight
View AI not as the machine itself, but as a tool used in conjunction with the machine. It's not about replacing human creativity but rather unlocking bandwidth for higher-level tasks.
When I engage in this AI-assisted workflow, I experience a flow state. I'm still fully engaged, thinking about how to structure prompts and approach the task at a higher level than before. The time saved on mundane tasks frees me to focus on the more creative aspects of the work.
There are still plenty of mundane parts of my job that I look forward to having replaced by AI. With less time spent on the mechanics of content creation, I can spend more time thinking about my audience's real needs and the jobs they need to get done.