Above: Watch Dave Altizer’s introduction to Sora.
EOSHD’s Andrew Reid takes a look at OpenAI’s powerful new generative AI video tool, Sora.
This is big news, a big shock really. The announcement today of Sora by OpenAI raises a lot of questions, but also raises the bar for generative AI video creativity far beyond that currently offered by existing competitors.
In the normal world of video production, stock footage and online content creation, jobs are now on the line like never before, but which ones exactly? Are you in the firing line?
Here I will give you my honest opinion of OpenAI’s new technology and at the same time the cold hard facts about Sora and what it could do, creatively and economically.
Depending on who you listen to AI is either the end of the world or a big opportunity that creates more jobs than it takes away. I say this – Sora has the potential to do both extreme wrong, extreme good and everything in-between. The first thing to realise is that “sora” means sky in Japanese. It is literally a clear blue sky – anything you dream of can be envisioned with it.
Sora is also an important step on the way to Artificial General Intelligence (AGI) and AI’s understanding of the real world, physical objects, physics itself. It is a world simulator not just a two dimensional moving-image generation tool. What’s shocking is the sheer rate of progress towards AGI – it is extreme and we will be able to envision much more specific results with Sora before very long, all while giving prototype AGI models more and more detailed and complex understanding of the real human world, which feeds into the creation a kind of magic, and a kind of doomsday weapon.
Along the way, Sora is going to be giving more and more control back to us as well, as artists, cinematographers and directors. At the moment a lot of AI art lacks intentionality, direction and consistency. It’s also difficult to bend it to a very specific vision you might have for it, as that vision is trapped inside your head and we can currently only communicate with it through short natural language prompts which aren’t really detailed enough or descriptive enough for more tailor made imagery, or via existing images and video that it will copy the style of but not really know what to do with over a period of more than a few seconds. This is going to change.
Sora will also be able to more closely replicate real situations, real people, real locations and this makes for a very controversial aspect of the tech, which will be under the spotlight of regulators and the in-house “red team” at Open AI like never before.
Jobs in the firing line
I have made a list of which film and video industry positions are in the firing line of Sora in 2024, and which new jobs will be created by this technology as it advances.
You can see the two lists further down the page. I think the real threat to our jobs however is not so much about specific groups only, there’s an even bigger danger in the way AI is scaling, and the way it augments human intelligence – the sheer productivity that can create in one person. The scenario is that one person could be able to do the job of 20 others, or do 20 times the amount of work 20 times quicker with 20 times less expense.
Your job might go away not because you’re not relevant, but because you’re no longer in demand.
For sure, your job might change and be augmented with AI. Art itself might change (and this of all things is THE most difficult area to predict). I think nearly all film and video roles will be touched or changed by AI in some way.
So before the lists…
Let’s talk a little bit about timeframes.
I give Sora 24 months before it is making a big impact on commercial videography overall. Some areas will change much sooner (like stock footage licensing).
Towards the end of that 24 months, Sora will be a 2nd or 3rd generation product which has not just ironed out some of the imperfections of the first gen, but revolutionised itself in terms of capabilities several times over.
So we have around 2 years before a big correction, I think.
The biggest change is that generative AI content specialists will be more and more in demand. It is a good time to make yourself a reputation in generative AI simply by making the time to use it, post your results and above all stand out with the quality of your ideas, and your artistic taste vs the average Joe.
You see, what is happening today is about the early ramping up of machine augmentation of human intelligence. This isn’t so much about replacing people with virtual humans – it is far more about augmenting (some of) the existing ones with artificial intelligence.
If your own intelligence or job isn’t worth augmenting, that doesn’t bode well. If on the other-hand your artistic talents, judgement and taste are in fashion and aesthetically in demand, you have a good chance of exercising a high level of quality control over your tools, Sora included. You will be in demand.
In this respect it is the same for AI tools as it is for traditional filmmaking and videography. This is a danger for mediocre imaginations, because from a client’s perspective it would make sense to hire just 1 person who has augmented their already existing creativity with Sora, rather than have 1 person who isn’t creative in the first place + a crew of 6 toil for days with obsolete tools to give a worse result, at much greater expense, with longer gestation periods.
Before too long clients will turn to people who are more specialised in generative AI than in traditional videography, and hire them instead of a larger crew of X number of people, with X number of cameras.
Partly however I see it dependant somewhat on how much reality they want – things like interviews, where there might be ethical problems in doing this generatively, even if it’s scripted, with the same end result, be the process real or imaginary.
Here are the roles in the video and filmmaking industry which I think are most under threat from AI video in the short term:
Most:
- Stock footage for adverts
- Medium sized commercial videography studios and crews
- Videography rental shops of all sizes
- Narrative indie filmmakers
- Narrative CGI renderers and animators
- Smaller scale videographers who specialise in web content and advertising
In the initial 2024 stages, high quality AI generated video is all about stock. If you are a client who is happily using existing stock footage – albeit real – and can get the desired result with it and calling it a wrap after a quick edit, you will be able to use Sora for that instead, so soon the real footage on stock libraries will be far less in demand, and will be crowded out by AI generated content. Regulators need to get on top of this quickly and URGENTLY make laws about labelling footage that is AI generated in advertising, political videos, and many other forms of video.
A few other observations: medium scale videography is next in the existential threat list, and I rate this to be more quickly and more urgently at risk compared to smaller scale freelancer work, and people just starting out. This is because most clients of the mid-tier production industry have the resources, awareness and need to hire in-house generative AI prompt engineers to do a content job previously done by independent video crews. A lot of these companies will really go for it with an in house crew of 2 or 3 staff, and they might also be the first to outsource to indie AI filmmaking companies as it will be more cost effective. In general they don’t demand the perfection that corporations want in terms of marketing, or the scale, or the A-list crews, nor can they afford the A-list celebrities that large corporations do. Equally, they’re generally going to be throwing more resources at AI than a small business or start-up world.
The rental shops with their shelves of unused cinema cameras are next at threat, simply because in a generative AI world, there’s fewer crews, and smaller crews. The need for mid-tier professional filmmaking gear isn’t restricted to just cameras but everything else that would otherwise be present on set, lights, video monitors, audio gear and so on.
Next up are indie filmmakers, because their content now has to compete with legions of AI simulated rival films without the same budgetary restrictions impacting the scope of their stories or the level of production value. Indie films (generally speaking) don’t have the saving grace of large scale production / Hollywood / Netflix – that of big name actors and the celebrity allure that insulates that level of production from AI content.
Art-house cinema already has trouble getting noticed by a mass-audience – it will far less noticed in the sea of more eye catching AI productions, thus less able to get funding or any kind of break-out traction via festivals or online platforms. Equally as unfortunately, the same probably applies to animated features at the indie level, especially CGI ones.
This mass redundancy of talent could apply higher up the food chain as well if AI can generate a particular style and rendering, the need to hire experts in CGI animation just isn’t there as before, even at the level of Disney / Pixar. This does not mean to say that indie filmmakers can’t shift to a different tool to make their scripts come to life, and embrace AI. Ethically speaking and in terms of principals, and in terms of enjoyment, artists might find this a hard pill to swallow.
On the flip side, AI has the potential to create a lot of new jobs and augment existing roles so that they stay relevant. Although AI is currently very easy to use and has a very low bar to entry, this does not mean that all clients in need of video content will simply all be able to do the AI part themselves without hiring specialists and professionals to get a better result.
I just fail to see that the economies of it will be as good to us as traditional videography and filmmaking was and is. Because with tools like Sora, you can make many more moving images, more quickly, with hardly any physical effort, without a crew, just one person by themselves, a brute force production engine.
Whereas a client might hire 5 or 10 people in a month to get the shots they need, now it might end up being one or two, and perhaps none if you are editing together AI stock footage.
When it comes to much larger scale production, the top-end, I think this will be impacted last and they will employ only the best in AI art, by which time there will be some very successful generative AI studios out there employing thousands of people. A big job creation area.
If you’re on the top 5 list above you might want to consider learning as much as possible about these growth areas:
- Generative AI content engineers, specialising in specific tools (like Sora)
- “Red team” roles involved in the moderation and testing of new AI models
- AI regulatory jobs at larger organisations and in government
- Generative AI artists and filmmakers with a distinct voice and track record of great material
- Generative AI artistic directors and project managers
- Writers, conceptual artists, musicians – real scripts, real stories, real music and real human imagination still has an appeal simply for the fact it is REAL lived experience and imagination
- Tutorial creators and social media influencers in the AI video world
- Teachers in AI at institutions and universities, specialising in AI video and photography
- Documentary and news shooters, those who capture REALITY and have no need to resort to AI
- Big name videographers and filmmakers who have the funding, name recognition and existing audience to continue the traditional way
Too accessible?
There’s no denying that the bar to entry and the ease at which a DIY approach can be taken with AI filmmaking spells trouble. It might leave a lot less room for professionals and highly skilled roles in the lower-end of production.
If a small business wants a simple advertisement, it just seems like a no brainer to feed their brand identity into Sora with a short prompt describing a few scene, a common thread, a slogan and out pops an instant result, as many iterations as you want, all for something like $50 per month.
With very high production values.
And if that doesn’t quite work perfectly now, it will in 12 to 24 months.
For them, the costs and complexity of shooting a real-life video that looks as good as AI suddenly doesn’t make sense. So they either settle for something that’s real but paired back and more amateurish, or waste money needlessly. Neither option seems to appeal to many clients, does it?
This won’t be the high-end content producers that lose out, because these smaller fish aren’t their clients in the first place.
The most concerning general trend here for me is the economies of it all, and the far smaller headcount required to do the same job with AI vs a traditional day of filming.
Many filmmakers and commercial videographers will be undercut in a big way by OpenAI’s technology. This is doubly insulting as a lot of their content was probably used for free and without permission, in training the massive models underpinning Sora’s tech.
Filmmakers and videographers now have a much tougher sell to clients in asking them to pay as much for an AI piece that took an hour to make as something which until last Thursday took a day of shooting and a week of editing. In order to get the same revenue as before you’d need to dramatically expand the number of clients you take on, leading to more stress and less job satisfaction.
Then there’s the other side of this brave new world – The stubborn appeal of reality. People do still respond to real, authentic stories and characters, real people, real actors, real personalities. Artists enjoy the process. Actors enjoy acting. Cinematographers enjoy being behind a camera.
You cannot get the same buzz from typing a few words into a chatbot.
You can make it resemble real people and events, you can use it to augment reality and real footage – like CGI and green screen do today – but this is only going to get ever more controversial as the misuse of the technology rises and more people are harmed by deep-fakery.
No matter how photo-realistic in style Sora may be, the fact remains that the content is synthetic, plastic, faked.
When the lines get blurred between reality and synthesised video, there is one big advantage real people in video and film still have over companies like Open AI and the legions of prompt engineers.
They’re human.
