The Development of AI Is a Violent Revolution
- No.
- 161
- Date
- Read
- 5 min
- Words
- 1,129
Technology is never neutral. It always changes the relationship between people and the world in some direction. —Neil Postman, Amusing Ourselves to Death
Ever since the arrival of the supreme, mightiest, kindest, most beautiful, most benevolent, most righteous, most sincere, all-capable, everywhere-present, all-knowing, all-embracing, all-having, all-doing, utterly sovereign, omniscient and omnipotent AI Coding tools, today another AI product has stunned me: ChatGPT's newest image generation feature.
Every time this happens, I feel more strongly that the development of AI is a violent revolution.
Over the past twenty years, plenty of industries were hyped the way AI is. In the end, none of them broke out with real revolutionary force, and they disappeared into the long river of history: the "metaverse," "NFTs," "blockchain," and so on.
AI is a concept whose first form appeared in the 1950s. After ChatGPT burst onto the scene in 2022 and set off the AI boom, three years have passed, and instead of fading like those earlier crazes, AI keeps advancing, and its revolutionary nature keeps getting clearer.
So what makes AI revolutionary? In my view, AI is revolutionary because it's a tool.
To be more precise, its usefulness as a tool is revolutionary because it's so overwhelmingly useful.
Overwhelming Usefulness, No Longer Built for One Skill
At first, AI was just a chat model. Its use as a tool was almost entirely conversation. Compared with Baidu or Google, it was better at summarizing, but its use was still fairly narrow, limited to things like searching and summing up.
Later, more AI products appeared beyond chat. NotebookLM lets you build a knowledge base beyond anything you'd imagine. You can use it for knowledge bases, mind maps, slide decks, even podcasts. One product multiplied what AI could do as a tool.
Then came "Vibe Coding." You can write code with AI? What does that mean? It means endless possibilities.
While AI was learning to write code, it also learned to generate images and video. At first the output was mechanical. Later it gradually developed taste, and even without detailed prompts it could produce beautiful images.
Making graphics used to mean learning Photoshop. Then it became Gaoding Design, where you make graphics right in your browser. Now it's AI, and one sentence does the job.
An All-Rounder Who Defies the Age
If AI were a person, it would be an oddball in modern education, because it wasn't trained to be an expert or talent in one area. It's an all-rounder.
There have been many all-rounders in human history. The most famous is probably Leonardo da Vinci: painter, sculptor, architect, musician, scientist, mathematician, engineer, inventor, anatomist, geologist, botanist, and writer.
Examples like that are close to godlike in human history. Since the Industrial Revolution, nearly every education system in the world has been training people to become "a cog in the machine," someone who's highly useful in one field. An all-rounder is more like a myth. That's why the heroes of web novels usually master the skills and tricks of every school. Being able to do everything looks like magic in itself.
So I think AI, at its core, runs against the age. It's too easy to set in motion, and it can do too much. Humans exploring what AI can do have already fallen far behind how fast AI is developing.
Endless Possibilities: What I've Done with AI
That's what AI is like. Using it is like having an all-capable partner. In a world built on AI, people and tools can create endless possibilities.
So what have I done with AI so far?
- Built my blog from scratch. I mean truly from scratch, and not a blog managed on a third-party platform, but a local Hugo-based blog that takes programming. With GitHub making things easier, I used AI to redesign the blog's style and theme.
- Developed a "Dialectical Materialism Q&A" app. It's what I've poured myself into lately, and what I'm proudest of. I kept feeding the AI my ideas and constraints, and in the end it became a real product.
Those are the two projects I've taken furthest so far and am happiest to show.
Beyond those, AI is now deeply involved in my life, writing, and work:
- Everyday questions in daily life
- In-depth conversations about writing ideas
- Checking my posts after I write them
- Making post covers (including this one)
- Making slide decks at work
The Giant Combine Harvester
I used to wonder whether AI's potential as a tool would be limited by its users, that is, by people. Does our understanding of a tool, and our relationship with it, set the ceiling on productivity?
It made me think of the harvesters I saw back in my hometown as a kid.
In the countryside, there are two kinds of land: dry fields and paddies. Dry fields are for vegetables, and paddies are for rice. The difference is that dry fields are fairly dry, while paddies are usually wet.
To work a dry field, people used to loosen the soil one hoe stroke at a time. By the time I was a kid, there were already small tillers driven by diesel engines. On one side of the machine was a pull cord. To start it, you yanked it hard a few times, and with a rumbling roar the machine came to life. That was the sound of productive forces.
In the paddies, the oldest way to harvest rice was to cut it by the handful, then beat each handful against the ground to knock the grain loose.
Later came small threshers. Like the tillers, they rumbled and roared. You cut the rice and fed it into the machine, and metal parts inside knocked the grain off.
The most advanced of all were the big combine harvesters. These machines were huge, about the size of a bulldozer.
They'd roll into the field to harvest the rice and sweep through it like wind scattering clouds, clean and quick. The front swallowed the rice plants, and almost at the same moment, the back end spat out heaps of grain.
It seemed that the stronger the tool, the smaller the role of people. Even with small threshers, harvesting rice still took a lot of hands. With a big combine harvester, it took little more than one driver taking the machine into the field.
But a big harvester like that still needs a driver.
Yes, a driver. As powerful as AI keeps getting, I think its results always end up acting on people. And since it acts on people, it should be governed by people.
Like the driver: however enormous the harvester is, it still has to turn where the steering wheel turns.
AI should be the same. So should the way AI develops and the way it's kept in check.