AI: We're Not Just Prompts!

AI’s trajectory isn’t just upward—it’s curving ever steeper. From DeepMind’s groundbreaking models to Flow’s democratization of filmmaking,

People are becoming used to how quickly AI technology improves.

Breakneck doesn’t even seem adequate to explain the scale of the movements. Because it isn’t just about the rate of change – even the rate of change of the rate of change is accelerating ... and the result is exponential progress.

Here is a simple example. Remember when you mocked AI-generated videos on social media for obvious flaws (e.g., six fingers, unnatural blinking or movement, etc.). Over the past few months, AI media quality has improved so much that spotting fakes is now difficult, even for tech-savvy people.

 

Well, we just took another giant leap.

 

This week, Google’s DeepMind unit released three new core AI models: Imagen for image generation,  Lyria for music generation, and Veo 3 for video generation.

It only takes a quick look at Veo 3 to realize it represents a significant breakthrough in delivering astonishingly realistic videos.

I’m only including two examples here ... but I went down the rabbit hole and came away very impressed.

Take a look – Everything in the clip below may be fake, but the AI is real.

 

via Jerrod Lew

The era of effortless, hyper-real content has arrived.

 

One of the big takeaways from tools like this is that you no longer need content creation talent other than your ideas.

An example of this comes from Google’s new AI filmmaking tool, Flow. 

What Is Flow?

What if creating professional-grade videos required no cameras, no crew, and no weeks of editing?

 

Flow can imagine and create videos just from your ideas. Kind of like telling a friend a story and having them draw or act it out instantly.

How Does It Work?

Think of Flow as a giant box of movie Legos. You can bring your own pieces (like pictures or clips) or ask Flow to make new pieces for you. Then, you snap them together to build scenes and clips that look like real movies.

Why Is This Cool?

It is becoming easier for almost anyone to create the type of content that only a specialist could produce before. The tool makes it easy in these three ways.

1. Consistent: The videos stick together well, so your story doesn’t jump around confusingly.
2. Seamless: It’s easy to add or change things without breaking the flow.
3. Cinematic: The videos look high-quality — like something you’d see on TV or in theaters.

If you want to play with it, it’s available to Google Ultra subscribers through the Gemini app and Google Labs. 

Ok, but what can it do?

Redefining “Real”

Don’t skip this next part. It’s what gave me the idea for the post.

To set the stage, imagine you’re watching a video of a person talking. Typically, you think, “This is real — someone actually stood in front of a camera and spoke.” But now computers can make a video that looks and sounds so real, you can’t tell it’s fake.

 

Anyway, this week, I saw a cool video on social media. At first, I thought it was cool simply because of the idea it expressed. But the video gets even more interesting when you realize how it was created.

 

“Prompt Theory” is a mind-bending exploration of artificial intelligence brought to life. The premise examines what happens when AI-generated characters refuse to believe they’re not real. From stunning visuals to synced audio, this video showcases AI’s new immersive storytelling power while examining some pretty trippy concepts.

 

Hashem Al-Ghaili via X

I predict you will see a massive influx of AI-generated content flooding social media using tools like this. 

Meanwhile, digital “people” with likenesses and internal objectives are increasingly going to become persistent and gain the ability to influence our world. This is inevitable. Yet, it’s still a little disorienting to think about.

 

As digital agents gain persistence and purpose, we face profound questions about reality, ethics, and human creativity.

 

And that is only the beginning!

 

Perhaps we are living in a simulation?