It’s been practically 24 hours since Google’s I/O 2025 keynote, and I’ve virtually extricated myself from the deluge of bulletins. I’m advised there’s a sky on the market with breathable air, however I’ll report again on the veracity of these claims as soon as I’ve seen it for myself. Quite a bit transpired yesterday, but additionally, in a means, little or no: there was AI for fanatics, AI for glasses, and a brand new AI with the potential to exacerbate your physique dysmorphia. What I’m saying is there was one factor (AI), however an absolute s***load of it. Whereas a few of that AI is (theoretically) sensible stuff, like a brand new mode for AI search or superior analysis fashions for heavy number-crunching, that’s not the AI that caught my eye essentially. For me, it was all about Stream.
Stream is a brand new product that Google is looking an “AI filmmaking software,” and it combines Veo 3, the corporate’s newest video technology mannequin, Imagen, its text-to-image mannequin, and Gemini, Google’s all-purpose massive language mannequin, which powers its ChatGPT competitor. With that holy trinity, you possibly can create totally generated movies from scratch—and never simply the video half, both, however sound too. One of the fascinating features of Stream is that, because of text-to-audio talents, it is truly fairly complete. Right here’s a cursed clip of a speaking muffin to get a way of what I imply.
My first Veo 3 gen
> a video with dialogue of two muffins whereas baking in an over, the primary muffin says “I am unable to consider this Veo 3 factor can do dialogue now!”, the second muffin says “AAAAH, a speaking muffin!” pic.twitter.com/VA2VUZF8sS
— fofr (@fofrAI) Might 20, 2025
And it’s not simply audio that makes Stream distinctive, it’s the software’s means to increase and manipulate scenes via digital camera controls (selecting angles, movement, and views) and continuity. “When you’ve created a topic or a scene, you possibly can combine those self same components into totally different clips and scenes with consistency. Or you should utilize a scene picture to start out a brand new shot,” says Google. Meaning, in idea, you possibly can actually chunk off one thing scene by scene and sew these parts collectively for an extended remaining creation.
The factor about Stream is that, on one hand, it’s type of magical. In simply a few years, we’ve gone from Will Smith consuming spaghetti to having the ability to craft whole-ass animated film scenes utilizing simply text-to-video/audio. Nonetheless you’re feeling about AI, that’s objectively a formidable technical feat. Alternatively, although, the philosophy behind instruments like Stream is nothing in need of bleak. In idea, having the ability to conjure up a film utilizing just a few textual content inputs and a laptop computer is a watershed second for motion pictures, however in apply, I’m undecided it would play out that means. If I’m to let my pessimistic mind take the wheel right here, I believe it’d result in much more of one thing, and artwork isn’t that factor—I’m speaking about AI slop.
As a lot as making visible artwork within the trendy age has benefited from the infinitely accessible digital toolset we now get pleasure from, these instruments (like Photoshop or Last Lower, for instance) are simply that: instruments. They aren’t there to make something for you, they’re there to take what you’ve made together with your flesh laptop (your huge human mind with all of its fold-y goodness) and understand the imaginative and prescient. I’m right here to say that’s a very good factor. A lot about artwork has nothing to do with type or type however about what it communicates from one human to a different. A communication from me to you. And the issue with producing video like Stream from begin to end—from thought to creation to refinement—is that it eliminates that human aspect solely. To be frank, that sucks, and I don’t learn about you, but when I had been able to being entertained or enriched by a robotic, I’d nonetheless be speaking to Smarter Youngster on AIM.
To be truthful, I believe Google does see Stream as an augmentative software and never a alternative, no less than to an extent. It’s even partnered with some actual filmmakers to point out you that its AI is worthy of your critical inventive endeavors. I don’t doubt that there shall be some who discover it genuinely helpful, too, and what they make may even rise to the extent of artwork. However principally I predict Stream shall be way more common with those that would slightly not hassle with ideating or creating in any respect, and a filmmaker that sort of individual isn’t. I assume we’ll see a technique or one other quickly sufficient since Stream is accessible by way of Google Labs proper now. Name me a pessimist, or a Luddite, or completely missing creativeness, however no matter you name me, simply be sure to considered it your self. Belief me, life is extra thrilling that means.
