Recently, I was part of an email conversation between Clayton Moore and Philip Hodgetts regarding the impact of AI (machine learning) on media production. I found their comments intriguing, so I got their permission to share this with you.
CLAYTON MOORE
I can’t remember, it may have been 2017 or maybe earlier. Philip Hodgetts talked about the high end of media production, the low end and how, at that time in his opinion the middle of the market was disappearing.
The new Roland AeroCaster, a $249 system which handles up to four cameras, four mics and live streaming in one small control panel using wireless iPhones and iPads made me think of his comment since he pretty much described this setup.
While the product was released over a year ago, I remember him talking about phones and iPad switching. Your “multi-camera stream in a backpack” is how he put it (as I recall).
Clients like Government or larger corporate types like to see rigs that look more “professional” of course, even though it makes no difference in the quality of the product where streaming is concerned.
Also, I saw a post Philip put up in Dec of 2022. Id love to ask him for his observations on “live captioning” which we use court reporter certified humans for right now. (I’d ask Philip directly, but I don’t know how to contact him.)
LARRY ADDS
I sent Clayton’s comment to Philip to get his reaction.
PHILIP HODGETTS
Hi, Clayton, and thank you Larry for passing this email on.
Reflecting on my older thoughts over the weekend, I realized that yes, I still believe the middle is disappearing. The business that I ran in Australia was very much based on a shortage of supply – quality video camera, editing equipment (pre-digital) and any sort of titling or graphics capability, along with the knowledge to use them, was hard for most people to get.
Therefore those of us who could supply them, and creative talent, had a business. Now I carry around in my pocket a camera with a quality I could only dream about in the 1980’s, with digital editing tools on the same device that outperform my Media 100 units by about 10:1 and it’s still in my pocket. Along with the means of direct access to distribution channels like Facebook, YouTube, Instagram, etc. There is no longer that scarcity that I was, in essence, selling in that business.
We’ve had at least one person turn up at LACPUG (user group) meetings with pamphlets advertising $150 shoot day, and $150 for editing. Clearly not a viable pricing for someone living independently of their parents!
Media 100 reached about 60,000 to perhaps 80,000 units sold, on the premise of “kill the editor”. Ahead of their time, they were preaching a democratized market that really happened with Final Cut Pro. (In a comparison I did once, there was a lot of overlap between John Molinari’s discourse about the vision for M100, and Steve Jobs’ introduction to iMovie many years later.)0
If you want access to high quality video shooting and editing, you can have it with a smartphone. The gatekeepers of access and quality no longer exist.
Now add to it, “Artificial Intelligence” (Machine Learning in reality but I’ll use the more common, but less accurate term) empowers individuals to do so much more creatively, at the cost of taking away that very career I profited from so many years ago! Truth is, my the early 1990’s I was seeing how quickly consumer video was evolving and realized that all I ultimately had to sell was the creativity and modified my business accordingly. AI takes that ‘promise’ even further.
Take Synthesia.io for example. Synthesia provides avatars for corporate and educational content. Licensed video is processed to be able to be used to generate any “talking head” video from text typed into a browser. A few months back they claimed over 6,000 clients. On the positive side, they’re so much easier to use that I have used them exclusively for Lumberjack System Help videos for the last two years. But on the negative side, that’s 6,000 people not shooting talking head, not doing the makeup, not preparing sets or locations, etc.
The voices are synthetic – from AWS for Synthesia, as are (my estimate) about 50% of all voice overs. See ElevenLabs! I know at least two well-known trainers in the post/video market (not Larry J. of course) who are considering using clones of their voice for voice over rather than their own human reads. One because the cloned version doesn’t mutter, um, er and repeat as real people do, and the other to avoid having to re-record sections to do a simple voiceover pickup in a training video.. Faster for his editor than waiting a couple of days for the trainer to get the pickup recorded.
RunwayML’s AI toolset gives me access to a lot of the visual effects capability of high end movies, at a very reasonable monthly membership.
As an individual creative, I am living in a golden age. As a small video production company I’m pretty screwed mid-term! Along with the camera tools and editing tools, there is a new crop of editing software that “automates” parts of the process, heading mid-term to “templatization.” MotionVFX has already started down that road with one of their apps.
Meanwhile, the high end making feature films and television hasn’t changed that much, and with the significant amount of money involved, will evolve much more slowly. Although if the AMPT had their way, extra work would all go to AI, at least the bit that hasn’t already with tools like Massive.
As Zach Arnold postulates, this is going to herald a return to the generalist, rather than specialists.
As for 2022’s “Fast Good and Free” we’ve seen free transcription become the norm. The cost of imagery and even short animations is going to approach zero with tools like MidJourney, Stable Diffusion et al. We’re almost finished working on a new app that brings Stable Diffusion to the local desktop with integration to FCP Projects and Premeire Pro’s Sequences for free, custom imagery on demand. The same will happen with short animations from text. Largely the domain of RunwayML version 2, but coming from others as well.
The cost of very expensive content creation tools is going to become close to zero.
Cheers, Philip.
CLAYTON REPLIES
Great thoughts Philip.
I saw rates for shooting go down years before COVID. I did DP work for a friend who worked back in the day for NBC NEWS.
For experienced skilled shooters I saw rates go from $750 a day, to $450 a day. From there it just went to not getting calls for work. If you charge a flat fee of $350 period it’s a matter of getting enough half or quarter days to make up for full days… Even then, it’s crazy.
As to the disruptive aspect of technology, I’ll never forget I was in a conversation with Russ Solomon (founder of Tower Records) at a reception a number of years ago, after retail record stores got hammered by digital. When he found out I worked for Apple, his whole countenance toward me changed.
My friend Adam Wilt recounted a story from NAB about twenty years ago where he saw long faces at the Media 100 booth. Final Cut Studio was just released, which then cost $1,300. They told him: “How do you compete with free?”
On a discussion board after the release of the very first iMac DV someone was being a bit condescending about how iMovie was not all that, and not professional.
I chimed in saying “You’re missing the point as someone (I was alluding to him) who just paid well over $5,000 for a basic semi-pro desktop editor. Apple just shipped a digital video NLE for free on a $1,200 consumer computer.” I went on, “that means a PC system for half what you paid won’t be that far down the road.”
All because of Apple’s digital video revolution. (However, I did not disclose I was the support product champion for iMovie 1.0 at the time. 😉 )
Captions – as they pertain to live steaming work, my human captioners say cadence and the inconsistent volume of people, along with accents, make automated systems achieving 98% accuracy very hard to do.
Non-technical people in the entertainment business and politics at this stage of the game, have pretty much no clue about AI and all its ramifications to make important decisions.
But they are headed in that direction. Keep an eye on this report commissioned by the California legislature to find benign ways to use AI in the State of California.
LARRY ADDS
Thanks, Clayton and Philip, for your thoughts. We live in interesting times. Feel free to share your comments below.
2 Responses to The Impact of AI on Media Production & Profits
I let Chat GPT write my response for me (just kidding, that thing is horrible).
I am/was a generalist my entire career, a do-it-all guy from beginning to end, making commercials, PSAs, documentaries, events, news, public affairs, oral history recordings, and training, oh so much training…. It served me well for over three decades, but I feel I got in and got out of the biz just at the right times in history to enjoy the easier tech of analog transitioning to digital, and was able to enjoy some A.I. tools’ help here at the end, but not get run out of business by it, specifically.
A.I. isn’t the only culprit: fast internet was the asteroid dooming our dinosaur-ruled business sector. When you have fast internet, your pool of possible editing contractors goes from five, very experienced, expensive, and well-equipped and established guys in Milwaukee, to five thousand guys and gals across the planet, time zones away, from Miami to Mumbai to Mozambique, working while you sleep, living in huts on pennies a day with no medical or other benefits. Retired factory workers in the US know the feeling of having their jobs off-shored. I knew my planned retirement hobby income doing home editing was destroyed, once Fiverr and similar services (enabled by fast internet) came on the scene with blowout pricing for editing and related services, and that was before A.I. was a thing. I can’t afford to turn my machine on, as a business, for just five bucks an hour… but a lot of those people, can. And do. Much of their work is junk. But not all. I still edit, but just for myself and a few friends. I can’t make revenue with it, unless I can monetize the viewing of my finished product on youtube. (I’m working on that, but it’s slow going for the content I want to make).
I used to work in government video, and in my shop, making those talking-head training videos was a good third or more of the business, and I did it with skill and taste and craft, and clients loved the level of care I put into the work, the viewers appreciated it, but, had we been a for-profit institution, instead of a free internal service, the clients would have left us in droves, once they discovered Synthesia, because we had a lot of friction in our administration regarding booking and prioritizing jobs for these clients. They never got the same treatment and fast service as the Governor or the Legislature, though that was our charter. Our bosses only cared about pleasing the most powerful clients and the “little guys” in the various agencies, boards and commissions had to take a back seat and just accept it. That’s a management/administration problem.
Now with Synthesia or something like it, the clients can skip the line, and the bad priorities, and non-returned phone calls of my ex-managers, and have a web service just do what they want, on-demand, on any timetable they need.
The demos I’ve seen of that kind of “training” are generally terrible from an Instructional Design POV; because the company providing the service relies on the clients to really understand how to communicate the information best… and a lot of mid-level office types never had those skills in the first place. You can recognize them by their powerpoint decks (shudders).
That was what I was for as a communications generalist: I could be their producer, fix their presentations, teach them -how- to teach, give their messages form, a narrative flow, and make them watchable and -effective- thru experience, design and craft. Now the clients can click on a web interface, upload bullet points and call it a day. That should make my ex-manager very happy; losing the clients that she hated servicing anyway. But in the larger scheme of things, it’s a loss on many levels, because it made my shop less and less strategically useful and more open to individual whims of an administration. I wont be surprised, the day they get phased out as over-specialized and irrelevant, replaced by A.I. But A.I. isn’t the only reason it will happen.
Will the storytelling be done as well, when you just hand off a script to an A.I.? Likely not, but these kinds of clients don’t always appreciate that distinction as long as they get what they ask for on schedule and they can tick off a box on a form that the requirements have been met. I pity the audiences that will have to rely on that programming; they aren’t getting the best. But they have no voice about it.
That middle range of projects Phil mentions is definitely under threat if not already gone. Market forces, enabled by these new tools, lead to business decisions that choose a less well made crafted product over the cheaper, faster, on-demand artificial one. That only leaves the high-end areas of the business as the last watering holes for revenue generation for us practitioners.
I’m remembering the scene in “Alien” when the severed head of Ashe the A.I. robot tells the doomed Nostromo crew: “I can’t lie to you about your chances…but you have my sympathies…”
Mark:
I think this is an eloquent, though sad, commentary.
It agrees with what I’m seeing. The easy money is going away, while the competition for the rest is growing.
Larry