Google’s Veo 3 AI Video Generator – What Is It & How Will It Change Video Creation?

If you’ve spent any time at all on the internet recently, you’ve probably been bombarded with AI-generated video content. From silly little clips of animals taking part in sporting activities to hyper-realistic and detailed promotional videos, we’re consuming a serious amount of AI-generated video. You might have even considered using it for your own business, or not even realised that what you are watching was AI-generated in the first place (but trust us, it is). 

The reliability of these kinds of AI tools has varied pretty significantly since their inception, but that recently changed. In May, Google launched Veo 3: its latest AI video generator, which, at first glance, goes above and beyond the capabilities of anything that has come before it. Naturally, this has got us thinking about what’s next in the ever-changing world of video, and wondering whether tools like Veo 3 are things we should be jumping on or approaching with a degree of caution.

What Is Veo 3?

Available to Google AI Ultra and AI Pro subscribers, Veo 3 is an AI video generation model capable of creating lifelike and high-definition videos from text and image prompts. What makes it notable compared to the tech that has come before it is the fact that it features native audio integration and can quickly generate videos with synchronised dialogue, ambient sounds, and background music. 

By generating audio within its own model (rather than relying on other tools to do so), the videos Veo 3 can create are incredibly realistic, often to the point where you might question whether the people or characters on display are human or not.

Take a look at this as an example:

Prompt: A medium shot frames an old sailor, his knitted blue sailor hat casting a shadow over his eyes, a thick grey beard obscuring his chin. He holds his pipe in one hand, gesturing with it towards the churning, grey sea beyond the ship’s railing. “This ocean, it’s a force, a wild, untamed might. And she commands your awe, with every breaking light”

Shared by Google themselves to promote their new model, this 8-second clip has been generated entirely by Veo 3. The sound is totally in sync and cinematic, the central figure is completely lifelike (wrinkles and all), and the clip leaves us with the lingering feeling that it’s not too dissimilar from anything a human can create.

Why Veo 3 Is Exciting

It’s all pretty exciting, right? That impossible-to-get promotional shot you’ve been after might now be possible. You’ll also no longer have to worry about having the right people in the right places if you’re creating video content. In theory, Veo 3 can do it all for you as long as you’ve fine-tuned your input and are willing to play around. Simply enter a prompt and receive a hyperrealistic video output, complete with the perfect framing and matching audio. 

Our Concerns

However, we’d be lying if we said we weren’t a little apprehensive. While we can’t dispute the fact that Veo 3 AI-generated video content is seriously impressive and could really streamline some video creation processes, it’s certainly not without its limitations. For starters, it’s only available to personal Google accounts at the moment (and at a hefty price tag of $249.99 per month with the “Google AI Ultra Plan). Accessibility issues are just the start, though, and we’ve already spotted a number of possible flaws that require some serious ironing out before Veo 3 can go truly mainstream. 

8 Second Clips Only

At the moment, Veo 3 can only generate clips of up to 8 seconds, making it currently more of a tool to simply have a play around with as opposed to generating any meaningful and usable product. In theory, you could enter prompts for individual cuts and edit them together to create something longer like an ad, but this would take you forever, with a number of prompts required for each clip before it generates exactly what you want.

Some Realism Flaws

We’re yet to see much wrong with the audio generated by Veo 3. However, the tool is still falling into some of the traps of the AI video generator models that have come before it in terms of total visual realism and accuracy. It can (and will) get things wrong, particularly written text.

The video below has been doing the rounds on social media and demonstrates this perfectly. Generated entirely by AI and hyper realistic, it’s been sending everyone into a bit of an existential crisis with the classic cries of “AI is going to take all of our jobs!”. However, the eagle-eyed will notice the video isn’t perfect. When the stereotypical biker-type male is being interviewed, you might spot that his waistcoat incorrectly reads ‘Hels Algels’ as opposed to ‘Hells Angels’. Yes, it’s a tiny detail, but it’s still wrong, and these kinds of errors have been commonplace across generative AI content for a while now. Veo 3 hasn’t solved them… yet.

Possibility Of Misinformation

This leads us to perhaps our biggest concern of all: the power tools like Veo 3 have to spread misinformation. Google has confirmed that watermarks are built into its AI-generated content, but whether anybody actually pays attention to this is another question. There’s a real possibility here for fake news to spread rapidly, for plagiarism to become rife, and for businesses and brands to make false claims (whether deliberately or not). If you can type any kind of prompt into Veo 3 and get a video clip back, you can start to see that some people might not use that power for good. 

In fact, plagiarism lawsuits related to AI image generation are already underway from the likes of Disney and Universal. While these lawsuits aren’t related to Veo 3 (it’s AI tool Midjourney in question here), they highlight how these kinds of AI models can quickly land users in some pretty hot waters. It’s pretty inevitable that the more Veo 3 is used, the more likely similar conflicts could come Google’s way. 

It’s weird to live in a world where you’re constantly considering whether things are real, and it’s important that we note that it’s not always a problem if something isn’t real. With AI video generation tools like Veo 3, creating video content is genuinely more accessible, and a lot of common frustrations or logistical nightmares will go away.

However, there’s no doubting that Veo 3 is a force to be reckoned with and not all of that force is good. The possibility for error is still there, and its output is limited. By all means, play around if you can, but don’t believe everything you see, and don’t forget that real human video creators (like us) still exist for good reason. 

In short: Google has launched Veo 3, a hyper-realistic AI video generator with synchronised audio capabilities. The videos they can create are undoubtedly impressive and certainly believable, but we’re still a little way away from tools like this making human video creation totally obsolete.