Has GenAI Lowered Our Standards?

OpenAI made a significant release this week; they launched their video generation model, Sora, to the public. Well it would have been available but the insane levels of demand caused OpenAI to throttle access to it. But, let's be clear, this is a seminal moment in the world of AI. Generative AI has advanced from text-to-text, text-to-image and now text-to-video. With video generation being incredibly more complex than text or static images. Social media and various commentators have not held back in sharing their thoughts on it, with the vast majority being in awe of Sora's capabilities. However, I think it still has a long way to go. This post on X perfectly sums up the challenges OpenAI still faces:

Play the video, and you'll see the gymnast is terribly formed. Admittedly, this is a very dynamic and complex scenario to produce, but it's simply not accurate. In any form of creativity, replicating people and human features is incredibly hard. Our own brains are so well attuned to observing our fellow people that we can spot anything amis from a mile off. You'll be very familiar with this if you've ever done any sketching or drawing. You might be okay with still life, but people, pah.

Who Will use Sora

So this begs the question: who is going to use Sora? Enterprise companies? Bootstrapped startups? Students? If any of the Fortune 500 were hoping they could save millions in production costs for their latest adverts, one look at the many examples of Sora already out there, and they'll be disappointed. Global companies invest billions to create well-honed brand images that they are proud of and go to great lengths to protect. Videos generated by Sora will not get anywhere their adverts or corporate comms, although as I write this, I'm aware that Coca-Cola has, in fact, already done this. In a previous post, I commented on the fact that the global drinks giant created three versions of its famous "The Holidays are Coming" advert purely using generative AI. However, the reception to these was lukewarm at best. Whilst it will have created some brand noise in the build up to Christmas, I doubt they'll be ditching traditional videos for their subsequent campaigns anytime soon.

Are Stock Companies Under Threat

That said, I think Sora poses a real threat to Stock Photo/Video companies such as Shutterstock. Need a video to get a message across in an internal presentation? Simply type out your prompt and paste in your video. If it's a small, regular meeting, I'm sure no one will care too much. If I look at myself creating Plain AI, I'm making exclusive use of ChatGPT to generate the images I use on the site. I don't think they're too bad and allow me to follow a certain brand style I'm aiming for that would probably be a lot harder with stock photos. But I'm a niche website, not a well known company.

What is 'Good Enough'

Back to the title of this article. Are we lowering our standards in place of convenience and cost savings? What are we now prepared to pass as 'acceptable'? What do we now deem as 'good enough' versus excellent, accurate and honest?

Whilst we must pause to acknowledge the incredible technological feat that Sora represents, we must also acknowledge there is still a long way to go until the output can truly replace current traditional methods.