PEDESTRIAN.TV has teamed up with 20th Century Studios for the launch of The Creator into cinemas.
The Creator – a movie set in a future where humans and artificial intelligence are at war – is out now in cinemas. With the film’s release (and just the general news about our incoming robot overlords) it’s got us thinking about our own relationships with AI.
Like every other person in a creative job, I’ve had hairy moments where I wonder about getting replaced by a robot. It couldn’t really do my job, could it? Surely not in my lifetime? What about when I have to go and argue with colleagues IRL — could a computer do that?
So many questions, and so little understanding about the tech. So, to try and calm my fears, I thought I’d outsource my job for a day to see how the technology stacks up. And, uh, spoiler alert: I’m not going anywhere for a looooong time…
Task #1: Coming up with headlines
This is a big part of my job. I pride myself on knowing the tone of my publication inside out. But does AI know it just as well? Let’s see.
Well, well, well. We’re already at strike one after trying to make a headline spicy (AKA our bread and butter here).
Hmmmm.
We also asked it to write a headline with a sci-fi angle (given the movie’s premise) and it spat out… this.
Beam Me Up, Pop Culture: The Cosmic Chronicles and Out-of-This-World Wonders of Sci-Fi Awesomeness!
Sci-Fi awesomeness?? Awesomeness?! Clearly this robot hasn’t ingested enough PEDESTRIAN.TV content because I’d get fired for approving that monstrosity.
I see where it’s coming from in the MAFS department, but it feels like my mum or an ageing millennial desperately trying to sound young and hip. It’s so enthusiastic! Loves an exclamation mark! Shook!
All in all, I’d give ChatGPT a 3/10 in the headlines department.
Task #2: Proofreading an article
Moving on to another big component of my job: reading and editing articles. I whacked one of my team’s articles into ChatGPT and asked it to edit it.
The result? It sent the piece back to me completely rewritten! This irrationally annoyed me because I had to look at them side by side to work out the differences. I demanded it tell me what edits it had actually made, which helped a smidge.
In saying that, the explanations were mainly things like “enhanced the sentence” or “improved the transition”. Useful.
I asked for more, and it gave me the same explanation. The real cherry on top was I spied some errors in there anyway.
Eye twitching, I concluded it would have been easier to proofread and give specific edits myself. A whopping 2/10 on this task.
Task #3: Responding to an email
I didn’t really know what else to ask ChatGPT to do because, so far, everything had been more time and effort than just doing it myself like usual. But I thought I’d give it one last crack with some boring admin. After picking one of my dreaded unread emails, I chucked one from a lovely colleague into the form and asked the robot to come up with my response.
Below is what it cooked up. (TBH, I added some redacted words in so that you, an actual human, can’t come and steal my job.)
I have never sounded this formal in my life. I think if I’d actually hit send on this my boss would have moseyed over to be like, “You all good???”. ChatGPT also couldn’t comprehend multiple respondents in the email thread and made me refer to myself in the third person.
Another flop. Probs another 3/10.
So, there you have it – humans: 1, robots: 0.
If you want to see more human vs. AI action, catch The Creator in cinemas now.