Lately, I’ve been thinking a lot about blacksmiths. Back in the days of horses and wagons, the blacksmith was a valuable member of society. But, with the advent of the automobile, blacksmiths faded away. They were replaced by something more efficient and, in most cases, better – factory-made parts and mechanics to install them.
While I’ve never had the skills to be a professional programmer, I’ve always liked programing. A couple of weeks ago, I started a new Python programming project. I wasn’t sure how to write it and I knew would have to learn a lot more about programming. But the deadline was flexible, the challenge intriguing, and I knew I could apply what I learned along the way.
As I described this to a friend, he asked why I didn’t let Claude Code do it for me? (For those that don’t know, Claude Code is an AI-powered coding assistant that helps developers build features, fix bugs, and automate development tasks.) Claude Code would be faster and easier.
True, I thought, but I wouldn’t learn anything. As I progressed, I kept writing, rewriting, and testing. It didn’t work, and didn’t work, and didn’t work. Two weeks went by when, suddenly, it worked! It was fast, flexible and perfectly met the need.
Even better, I understood every line in it. The sheer joy of creating this far outweighed the pain of learning how to do it. Claude Code would have been faster, but I would have learned nothing, could not have maintained it, and it would not have been better than what I created.
AI, today, already creates the majority of videos on social media. There is no doubt that within a few years, AI will create the majority of all the videos we watch. As you know, many of the videos we create today are the same as the ones we did yesterday, but with different content. AI is very good at iterative work.
The lone exceptions are live events – from news to weddings to sports – and even “live events” are challenged. Tools in today’s NLEs can record and edit multi-camera interviews based on who is talking. I’ve seen tools that automatically edit B-roll to match what the speakers are talking about. This spring at NAB, TVOne will showcase live event technology that simultaneously controls lighting, video playback, pyro, scenic automation, lasers, and audience wristbands from a single cue on one console. NBC used an AI version of Al Michael’s voice for event narration; they called it “A.I. Michaels.” Last fall, Tilly Norwood, an AI-generated character, applied to SAG-AFTRA for representation.
The videos that AI creates won’t be very good, but they will be “good enough.” YouTube is a classic example of “good enough” video. Good enough to attract a large audience, but not good enough to be good. Disposable, low cost, good enough videos will be the domain of AI… and clients on tight deadlines or budgets.
Perhaps regulation can help? In this political climate, meaningful regulation is highly unlikely. But, even if it were, regulation is effective at setting standards for job safety, product quality, or equal treatment, but regulations are not effective at determining staffing levels or locking in job security.
The world is changing and many us will need to cross-fade into other jobs. But not all of us and not all of the time. There will still be times where “good enough” isn’t good enough. Not all video is designed to be disposable.
Blacksmiths are still needed today. But not as many, and their craft is reserved for special projects.
Masters of the craft of visual story-telling with the skill to create new and compelling ideas will still be needed. But not as many as before. AI can “get the job done.” But it isn’t the same. Not for the creator, nor the people who watch it.
5 Responses to Will AI Impact Media Jobs? Absolutely Yes.
Larry, when you say that AI is making many of the videos on social media today and that it will be making all of them soon, do you mean it’s actually creating and posting the content or is it furnishing the methods by which the videos are created by humans? If it’s the latter, is there a parallel to be made around the switch from linear to non-linear editing? Only now the changes have to do with speed, process, flourishes and an earlier hand-off. But if it’s also content, then we’d all best become paid-up members of our blacksmithing associations; the ones that try to stop the spread of those smelly honking machines.
Steven:
From my understanding it is more than tools – which I have far less objection to – but actually creating the videos themselves.
I agree that AI tools that revolve around machine learning – masking, speech-to-text, color normalization, captioning, are highly useful because they keep the human in the loop. I use them happily myself in all three NLEs.
It’s the wave of AI-generated “slop,” which continues to grow, that I object to.
Larry
There’s a line in my favorite movie, The Man from Snowy River, where the young protagonist is asked his opinion of whether the mountains could be tamed. His answer: “I think you might sooner hold back the tide than tame the mountains.”
I think that’s unavoidably where we are with AI, but I think we do have an advantage over the blacksmiths, buggy makers, and auto plant workers displaced by technology in the past.
For one, thanks to those historical lessons, there’s more organized and intelligent resistance from the start to its ill effects, so that hopefully they’ll be managed better, earlier.
Two, on a related front, I think we’re at the stage with AI where the automobile was with the Models A & T—it’s been around a while, long enough to hint at where it could go, and now it’s beginning its charge into the sectors and jobs it can truly transform. That’s given the workers in those sectors time to consider what other avenues they have in front of them. But I do believe those avenues are going to be highly constricted, depending on the sector.
On that specific point, though, I think video content creators are in a somewhat better position than, say, secretaries and paralegals, so much of whose jobs are able to be automated. Video content and editing requires a level of creative judgment that I don’t think AI is going to be able to match, at least for a while, and in some cases I think at all. Certainly the AI slop you and I have both seen on YouTube bear that out at least for now. It’s uniformly terrible.
Even as it progresses, though, I think you’ll see a premium placed by consumers on “human-created content.” There’s already so much of a pushback already against AI-generated content that I think a real market will exist for simply knowing you’re watching and listening to content created by other humans.
That much I’m optimistic about.
Clint:
I appreciate your thoughts and hope that you are right.
Larry
I, too, think that there will be a huge desire on the client’s part for “human-created.” It will be very similar to the food industry advertising things that are “hand made,” or the Film Industry making “Practical Effects” as opposed to CGI. It’s things like that which many people will hanker for instead of just one giant AI swamp of content.