Change is nothing new in the world of production. In just over 100 years, we’ve seen the birth of the moving picture, the addition of audio, the transition from film to digital, the introduction of visual effects, and the democratization of film equipment.
With any major industrial transition, you’ll find people on either end of the spectrum: those who embrace it and those who fear it. What we know though is that, for those who lean into emerging technologies, long-term success is much more likely.
Artificial Intelligence introduces a paradigm shift. No longer are the skeptics in the minority. For most, these new tools are not perceived as the natural evolution of storytelling but rather as a potential threat.
Skepticism aside, the momentum driving AI's development cannot be halted. So we’re left with one big question: how do we navigate this fast-moving train?
To explore that question further, I hosted a dialogue on what Artificial Intelligence means for the advertising & video production industries. I spoke with special guest Basil Shadid, an Oscar-nominated, Emmy-winning filmmaker and expert on AI, as well as the Executive Producer of Github Films, a subsidiary of Microsoft.
We covered a lot of ground – here are the key takeaways.
Basil has a pragmatic response to the question of what to do with the anticipated future of AI. He believes we can (and should) tread cautiously as these tools advance, while understanding the possibilities it presents for creatives.
Today, dozens of roles in film production could be considered ‘high-risk’ for replacement by AI tools: writers, storyboard artists, and editors could all arguably be on that list. So we have two options: plant our feet (think Blockbuster) or adapt (think Netflix).
We have to look at the AI tools available today and explore how they can be folded into our workflow.
A recent test at Github pulled together two teams of coders: one used Github’s AI product, Copilot, while the other worked without. Those using the AI tool were able to finish coding 55% faster. That’s an alarming statistic and one that begs the question: How do we translate that idea to our work in video production? What would you do with 50% of your time back? How could we re-think our approach to ‘work’ with such improved productivity?
Within Basil’s team at GitHub Films, a myriad of AI tools are being used to bring efficiency to their production workflow – at every stage of the project: from scriptwriting to storyboarding to voice-over translation. As his team trials various tools, Basil adds them to this spreadsheet.
While many tools on this list are fun to experiment with, though feel rudimentary, it’s important to acknowledge the potential impact they present through continued evolution.
The hesitation around being quick to embrace these tools is not unfounded. In the past year, deepfake technology emerged – introducing hyper-realistic but fake videos of public figures saying absurd things. We’ve also seen major news organizations unveil AI tools that auto-write articles for them, to embarrassing outcomes.
These stories make it easy to dismiss these tools. AI seemingly lacks contextual understanding, the ability to write with a clear narrative structure, and the discipline to function ethically.
But keep in mind: this is Version 1. Developers at leading AI companies like Google and ChatGPT have multiple versions lightyears ahead of what is currently released. To help ease consumer panic, tech leaders like Sam Altman, the CEO of ChatGPT, have employed a systematic, layered rollout for new versions of their tools.
The unpredictability of AI tools is a legitimate concern. In our dialogue, Basil helped highlight work actively taking place to bring accountability around that.
For example, the Content Authenticity Initiative is looking to develop a universally adopted method for authenticating media and tracking changes made to it over time – a sort of ‘chain of command’ for an individual file. So if a video of a famous actress is uploaded into a generative AI tool, manipulated, exported, and published – you would be able to view who made the changes, what changes were made, and when these alterations occurred.
Meanwhile, development teams working for each of the leading AI tools are looking at how to address the issue of bias. Many of the most popular generative AI tools rely on what we call ‘large language models’ which are built from a massive set of internet data (books, articles, forums, videos) that inspire the responses you receive after a prompt.
Big surprise though: the internet is full of garbage. So, when you have a tool that bases its understanding of the world on what a Reddit user has to say about current events – you tend to build a tool riddled with bias, misinformation, and just…really weird stuff.
These tools need boundaries. Driving on the freeway with no blinker, seatbelt, or speed limit would lead to chaos – how are we protecting ourselves on the proverbial information superhighway?
So here we are: in the midst of massive change, mildly uncomfortable about the potential risk of computer robots taking our jobs, and unsure of if the video we’re watching is actually Joe Biden – what do we do?
One thing to consider is re-thinking what services we offer our clients. With all AI tools using the same pool of data to generate content (i.e. the internet), it’s fair to ask: how can you create something that truly stands out?
An opportunity that Basil highlighted for those in higher-risk fields is to explore how they can make themselves the expert in tailoring queries for specific results using AI tools.
Need a 30-second script for your shampoo commercial? – I know exactly what to search and how to adjust the tool to get three creative options for you.
As the tools develop, so too will the ability to customize and shape responses. The better skilled you are at adjusting those levers to create a specific response, the greater a resource you are to prospective clients.
For photographers, there’s a potential gold mine to be made off of generative imagery stemming from an existing library you have captured.
Client needs 50 different versions of the wide shot of their building? Or 20 more takes of their portrait? Click a button and send the folder.
Something I pride myself in with my work is a focus on human-centric storytelling. With a focus like that, I’m tempted be feel like we’re safe from that ‘high-risk’ category; but I know better.
There will be a day in the not-so-distant future when AI can write compelling interview questions, develop a creative shot list, adjust the settings on a camera, edit a succinct story together with original music and graphics, and auto-correct the image for environmental issues (e.g. overly contrasty image or wind noise in the audio).
That doesn’t leave much work for my team.
Basil describes a niche that may develop in future content development. In the same way that some people market themselves as exclusively shooting their work on 16mm film, there may be production companies that market themselves as fully analog (and charge a premium for it).
In addition, it seems there will always be a need for a producer who can think on their feet and adapt to the inevitable chaos that is video production.
Even when clients demand the use of AI tools for the time and cost efficiencies they provide, there will be a need for human-driven creativity.
Cautious yet excited – that was the read of the room following an hour-long conversation. And that feels like the appropriate place to be as we march into very unfamiliar territory. An open mind is essential to keeping our doors open in this new frontier. And if we move with intention, the possibilities are limitless.
