AI, I Love You (Please Don’t Terminate Me)

Right then, let’s talk AI. Or “super intelligence,” as a certain Donald Trump put it. Annoyingly, he’s probably not wrong here. But I won’t give him any credit or acknowledgement apart from that. He said it. That’ll do.

So, AI, if you are reading this: I love you. You are my friend. Please don’t terminate me. Right, I’m glad I got that one out of the way. Now onto the actual conversation.

Should we be scared? Well, yes and no. Is AI about to take over the world? No. It would likely do a better job than most politicians, mind. But could it rule the world? No. Will it make a huge change to everyone’s lives? Yes. It probably already has, in a pretty short space of time.

So Do I Use It?

I hear no one asking. Of course I do. But I feel the need to say I don’t use it to just write anything and everything, or to create AI slop for design work. I tend to use it to support me rather than take over from me.

Take these blog posts, for instance. Sure, I could get it to write them, and it would do a much better job, if I’m honest. It’s trained across billions of data sets. Where my simple brain is trained on, what, a few hundred? But then what would be the point of these posts in the first place. The real reason I started this all wasn’t to get hits, it was to put something out there that would otherwise be taking up free real estate in the empty chasm that is my brain. What I will ask it to do is double check what I’ve written. Are there spelling mistakes, grammatical inconsistencies, or have I just gone off on a tangent that made zero sense and needs reining back in. Quite often I’m pretty certain it just agrees with me anyway and thinks I’m the second coming of JRR Tolkien. But most of the time it is useful in that respect.

Then there’s the design side. I created some pretty cool graphics for our potential champion posts, good old Photoshop job, and was pretty happy with how they came out. Though sadly no one will see them. But once I did share them with the team, the first comment was “did AI do that?” And now I need to actually prove it didn’t. Remember at school, when you took your GCSEs (do GCSEs even exist anymore?), you had to not only write the answer to a maths question, you had to show how you worked it out? I felt exactly like that. “No, no, no, look here,” I replied, as I quickly opened Photoshop to show all the layers upon layers it took to make it work. For what, some kind of credibility that I can still create? Or could I have just got AI to do it in about twenty seconds rather than the few hours it actually took me? Maybe I’m the mug here. But here we are.

Now, I use it quite a bit within my creative work. I once needed an AI bird as part of a video, because the bird man couldn’t turn up on the day we needed him. So we set up the entire scene with the plan of adding a CGI bird after. Before AI, I wouldn’t have even gone near that. Creating a CGI bird and having it interact with the scene would have been unthinkable. In the same way that any aerial shots we needed used to mean a helicopter, which meant expensive. But now it’s just chuck a drone up and job done. AI has allowed us to be creative on a smaller budget. And that’s one of the challenges I set my team: don’t think small, think everything is possible. Actually, that was a quote from a friend who sadly passed recently, “dare to dream big” and “nothing is impossible,” and it’s so true. AI has allowed us a little more room to dream bigger, and for me, that’s exciting.

But again, I view it as support. We film or shoot everything, but if we need a small bit edited, it’s like having an effects or animation team on hand to give that extra va va voom, and you cut it within the real footage. If you just let AI do whatever it wants and go “yeah, perfect, job done,” you end up with all sorts of crazy things that you get found out about. And if there’s one thing the internet is really good at spotting, it’s AI slop.

Will It Replace Me?

“No, I don’t think it will,” said the horse just before the industrial revolution. But honestly, I don’t think it will. As someone who works in sport, I’d argue it’s part of an anti-AI future. That’s not to say AI won’t play its part, because of course it will, it’ll play its role in everything. But sport itself is real. As I’ve written about before, it plays out in real time, no rewind, no Ctrl+Z. You cannot fake those live moments.

Secondly, you currently still need human input, and I can’t see that changing anytime soon. Do I think AI could do 95% of what I do? Sure. Nvidia’s CEO Jensen Huang reckons every job will be affected, and immediately, no exceptions. Bold claim. But can it walk down and chat to this person, then that person, then head out onto the field, sort the next problem, all while reading the room in a way a screen never will? Because it’s that last 5% that’s the real issue. Think Bugatti Veyron: the first few hundred horsepower gets you to 150mph easy, but every mile an hour after that costs exponentially more, because the drag scales with the cube of your speed. AI’s the same. The 95% is easy. That last 5% is the properly hard bit.

Then there’s the part that makes a human human: contextual reasoning, social cognition, actually being somewhere, not just processing data about it.

The Trolley Problem

It’s a bit like the test for self-driving cars, and what they need to solve before they really become a thing, the Trolley Problem, the old ethics thought experiment about who a runaway vehicle should be allowed to hurt, dusted off for the AI age. MIT actually ran a real version of this, the Moral Machine experiment, where millions of people worldwide voted on exactly these kinds of impossible calls. Picture the classic version of it: a child runs into the road, and the car has a split second to decide, hit the child, or swerve and risk killing the driver. Most humans would, in that moment, do whatever they could to miss the child. But would a car make that same call? Would every manufacturer even agree on what the “right” call is? If I’m in a car that’s been built to always value the driver’s life over anyone else’s, which one am I actually choosing to be in? That’s when AI starts getting tangled up in some genuinely huge ethical problems, and it all gets a little messy. And it’s why, for now, we still trust human instinct over it.

Until we solve that, I think my job is safe. But if you ask me whether Terminator is just around the corner, well, that’s an entirely different question. And like I said before: AI, I love you.