The AI video gold rush: A tale of hype, hope and hallucinated fingers
A.I.M. Newsletter Issue 11 - Lead Story (via LinkedIn)
A.I.M. NEWS - By CJ Knight - 12/20/24
The industry's latest fascination is a $200 monthly subscription to make slightly unsettling videos that your customers would rather not watch. Welcome to 2024's AI video revolution.
OpenAI and Google are duking it out in the next big AI arena: text-to-video, with Sora and Veo respectively promising to transform how we create visual content. The demos are impressive – if you ignore the occasional extra finger and that peculiar sheen that makes everything look like it was filmed inside a snow globe that survived a microwave accident.
But here's the sobering reality:
NielsenIQ just dropped a study to be fully unveiled at CES 2025 that should make every CMO pause before diving headfirst into the AI video pool. It turns out consumers find AI-generated ads "annoying," "boring," and "confusing." Not exactly the trifecta you're aiming for when building brand loyalty.
The human brain, that stubbornly analog device, apparently struggles to process these AI-generated visuals. While OpenAI's Sora can now generate 20-second clips in crisp 1080p (for the bargain price of a monthly used car payment), and Google's Veo 2 promises 4K resolution (for those on a waitlist), they're both solving for technical specifications while missing the human connection.
Just ask the Queensland Symphony Orchestra, whose AI-generated social media campaign had audiences counting fingers instead of buying tickets. Or consider the infamous AI fiasco a la Willy Wonka Experience, where dazzling AI-generated wonderland visuals led to refund demands when guests encountered a sparse warehouse and a dejected Oompa Loompa. Even customer service isn't immune – DPD's AI chatbot went rogue, cursing freely and writing poetry about DPD’s demise.
Ramon Melgarejo, president of Strategic Analytics & Insights at NIQ, puts it diplomatically: "Brands and agencies are innovating at a rapid pace." Translation: They're running so fast they forgot to check if anyone wants what they're chasing.
The research is clear: even high-quality AI-generated ads trigger weaker memory activation than traditional spots. Your expensively generated AI content might look polished, but it's bouncing off viewers' brains like a tennis ball off a brick wall. The “uncanny valley” isn't just a visceral human response to creepy or unnatural AI anymore – it's where your marketing budget goes to die.
This doesn't mean AI video is dead on arrival. The technology is advancing at a blistering pace.
OpenAI's new storyboard feature and Google's improved physics modeling show promise. But right now, these tools are better suited for pre-production: ideation, storyboarding, and rapid prototyping.
They're the sketch pad, not the final canvas.
The platforms are trying to address concerns. Both Sora and Veo come with built-in safeguards – watermarks, metadata, and strict content guidelines. Google claims Veo 2 hallucinates fewer unwanted details, which is a bit like saying your new car only occasionally drives itself into a lake.
As we move into 2025, the smart money isn't on replacing traditional video production – it's on augmenting it. Use AI to rapid-prototype concepts, generate test variations, or visualize storyboards. Success requires collaboration between AI tools and expert human marketers who know when to use them – and more importantly, when to rely on more traditional creative approaches.
But when it comes to your final cut? Maybe stick with tools that don't need to be reminded not to add extra fingers. The revolution isn't quite ready for prime time. But it's ready for pre-production.
###
