Artificially yours
Is there still a role for creative skills when you can pretty much generate anything you like using AI these days?
It's true, it's super-easy and sometimes indistinguishable from the real thing, but in my experience, media is best served by AI as a tool rather than a full content creator. I have listed some of my favourite uses of AI in my production below. Before getting to that, let's just appreciate Canva’s quite unbelievable attempt to create something for me last year..
I had a portrait image of a cute baby rhino from Chester Zoo and I wanted a landscape version. I couldn't find it online so I thought I would ask Canva to do it instead. After all, it claimed to be able to generate expansions on your images so it seemed like a relatively simple task.
First attempt: I drag the area where I want the image expanding to…
Ok, what is THAT?? Canva somehow took my baby rhino and thought I wanted a parent running alongside it. Only Canva also thought I wanted its parent to be a species of animal that doesn't exist (maybe part sheep, part anteater?)
2. Second attempt: I am going to help Canva out by not highlighting the area with the baby rhino in..
Right, I am beginning to think Canva has an obsession with this ridiculous creature now! It is smaller this time. It also appears to have placed its front legs inside some Winklepicker boots for some reason best not known.
3. Final attempt: I am clever, I have worked out the Canva's AI cannot get over the fact that there is an animal in the image and wants to give it a friend. So, I remove the baby rhino entirely. I will generate just more of the grass then add it back onto the original image. Sorted!
WHERE DID HE COME FROM??? I didn't specify I wanted anyone in the shot but Canva has not only made me a smiley man in a smart, branded polo shirt, but also the remnants of a lower third to announce his arrival! Thank you, Canva.
Where AI does actually help me:
As I found out, sometimes AI just cannot be relied upon and certainly shouldn't be used for things that are far more simply fixed (like just asking my contacts at the zoo for the original image!) However, sometimes it just makes sense to save huge amounts of time or to help out when the right tools for the job are not available to you.
Voiceovers - why on earth would I want to drive to a studio, have someone rehearse the script, record them on multiple takes, edit the resultant audio clips on my computer and then still potentially be disappointed with the results as the person was just not confident enough, sounded like they were reading, missed a key word or just couldn't pronounce something vital. In a few seconds, I can upload the script, choose a voice (gender, age, tone, dialect etc.) and have a really high quality voiceover. Things have really progressed in the past few months and the available voices are unbelievably realistic - no robotic monotones for me, thank you! Envato's AI voice gen is getting really good and doesn't cost anything on top of my usual subscription.
Captions - adding subtitles with the click of a button in Premiere Pro is sooooo handy. I never really considered it to be AI and while writing this blog it struck me that it must be, so here it is and it's high on my usage list. It isn't perfect, but it's waaaaay faster to edit auto-generated captions than
AI Upscaling - this is easily my no. 1 use of AI in my production! Especially when building client websites. Schools are the worst for this as the images available have probably been taken on a phone or iPad (or they are a screenshot!). Do I want to build them a lovely new site that has stunning design features and crisp text only to taint the whole experience with horrid, grainy images? Of course not. Without the option to just rock up and snap a whole load of new images, sometimes it is just quicker and easier to use an online upscaler to take the original and work its magic. The results are usually astounding! I tend to use https://imgupscaler.ai/ which has a really good free tool and takes seconds.
Background removal - another really common use of AI is just clicking a button in Photoshop, Adobe Express or Canva and having the software get rid of all that nonsense around the outside of your subject. I know how to do it manually and often I still have to get my hands dirty when the AI fails but on the whole it's pretty perfect and saves hours of work across a month.
Google Lens - I don't use it a lot, but if I need to find an original image using a small sample that a client has provided, it can prove to be really useful. I also used it to identify what I believed to be a False Widow spider I found under some old decking in the garden, but that has nothing to do with this theme! (FYI 2 different images of the same spider produced 2 different results from Google Lens..)
Images - stock images are still really useful and I use them a lot. However, very occasionally you just cannot find the right thing. Stock images are often exceptionally clean or corporate. If you want something less obviously stock, maybe AI can help out. Of course, this is all only really something I would do if totally unable to get the actual photo from a client or take one myself.
Video - it is rare, but I will very occasionally generate a video using AI. Only short snippets and only things that would otherwise completely outweigh the time, effort, travel and processing to do it properly.
Scripts - recently, a client asked if I knew someone who could take a page on their website and make it into a script. I could have tried it myself but that's not my area of expertise. I could have searched for someone who is an expert. As an experiment, I used ChatGPT to create a full script in a corporate presentation style and then to refine it for a shorter social reel without losing the key message. The client made about 3 small changes to the script, that was all.
Responsible use of AI:
Throwing yourself into AI needs to be tempered with a little bit of common sense and a lot of consideration for the environmental impacts it can have. That kind of processing power doesn't happen without some serious energy usage and so I try to balance my own deployment of it against the alternative.
For example, If I need a short video for a client website that doesn't need to reflect an actual place or real people, I will quite happily get AI to create me something. The alternative is that I have to drive to a client site, organise people to take time out of their day to be filmed, use my editing suite for creating a film from the footage and then render and export it.
What's next?
It's hard to see where this pathway will lead. It undoubtedly needs to be tamed before it gets completely out of control (does anybody trust any content they see on social media anymore?) but where it helps the creative process without taking it over, it's something I am happy to add to my toolkit.
